文字起こし 🔗
WEBVTT
00:00:15.679 --> 00:00:18.399
Welcome to artificially intelligent marketing,
00:00:18.519 --> 00:00:21.519
a weekly podcast where we stay on top of the
00:00:21.519 --> 00:00:24.820
latest trends, tips, and tools in the world of
00:00:24.820 --> 00:00:27.620
marketing AI, helping you get the best results
00:00:27.620 --> 00:00:30.780
from your marketing efforts. Now let's join our
00:00:30.780 --> 00:00:34.780
hosts, Paul Avery and Martin Broadhurst. Here
00:00:34.780 --> 00:00:37.140
we are again on Artificially Intelligent Marketing,
00:00:37.219 --> 00:00:40.960
and we're a second episode back and already know
00:00:40.960 --> 00:00:42.719
Martin Broadhurst, but that's all right because
00:00:42.719 --> 00:00:45.020
today we've brought in an upgrade. We've got
00:00:45.020 --> 00:00:46.979
my friend Harrison Wade from the life science
00:00:46.979 --> 00:00:49.659
industry. Harrison, how are you doing? Good,
00:00:49.700 --> 00:00:52.000
Paul. Great to be here. I'm so glad that you
00:00:52.000 --> 00:00:54.979
guys started this podcast up again. I think it
00:00:54.979 --> 00:00:57.340
was almost like two years ago or something. I
00:00:57.340 --> 00:00:59.770
remember listening to like pretty much every
00:00:59.770 --> 00:01:03.329
episode at the beginning and uh it was like oh
00:01:03.329 --> 00:01:05.909
it just disappeared for a while and i know we
00:01:05.909 --> 00:01:07.950
had chatted about you know why that was happening
00:01:07.950 --> 00:01:11.069
but uh which is obviously exciting but yeah i'm
00:01:11.069 --> 00:01:13.150
so glad it's back and obviously excited to be
00:01:13.150 --> 00:01:15.730
here yeah i love it why don't you tell the um
00:01:15.730 --> 00:01:17.769
the listeners a bit about yourself you know your
00:01:17.769 --> 00:01:21.219
story to date what you're doing now yeah Yeah,
00:01:21.239 --> 00:01:24.500
so actually, I fell into the life sciences later
00:01:24.500 --> 00:01:27.180
in my career. So I think most people in this
00:01:27.180 --> 00:01:29.700
industry started off probably in the lab and
00:01:29.700 --> 00:01:32.219
then maybe moved into commercial, into marketing
00:01:32.219 --> 00:01:34.879
later in their career. I actually started in
00:01:34.879 --> 00:01:36.959
software sales at the beginning of my career,
00:01:37.040 --> 00:01:40.540
selling to IT and then marketing, and then ultimately
00:01:40.540 --> 00:01:42.480
fell into the life sciences at a company called
00:01:42.480 --> 00:01:45.200
Synthego and joined relatively early to help
00:01:45.200 --> 00:01:47.920
build out the inside sales team, a lot of the
00:01:47.920 --> 00:01:50.450
systems processes, the sales training. implement
00:01:50.450 --> 00:01:54.510
functions there. And one of the things that we
00:01:54.510 --> 00:01:57.390
saw along the way was that a lot of reps were
00:01:57.390 --> 00:01:59.790
coming in. These really skilled reps would come
00:01:59.790 --> 00:02:01.349
in and say, wow, you guys have some of the best
00:02:01.349 --> 00:02:03.689
tools, some of the best tech, some of the best
00:02:03.689 --> 00:02:06.310
training that we've seen in the industry. I was
00:02:06.310 --> 00:02:07.790
like, well, that's interesting. I didn't think
00:02:07.790 --> 00:02:09.669
it was that cool. Because I think coming from
00:02:09.669 --> 00:02:16.009
the software world, a lot of the stuff was normal.
00:02:16.129 --> 00:02:19.310
It was very normalized. But I think even still
00:02:19.310 --> 00:02:21.810
today in the life sciences, like there are many
00:02:21.810 --> 00:02:24.810
companies that aren't even adopting technology
00:02:24.810 --> 00:02:28.629
that was used 10 years ago and maybe even more
00:02:28.629 --> 00:02:32.909
than that 15 years ago. And so, yeah, it was
00:02:32.909 --> 00:02:35.949
kind of like this opportunity where I moved to
00:02:35.949 --> 00:02:39.389
London. I was working with my co -founder, Nick,
00:02:39.590 --> 00:02:42.550
who I met at Synthago. And we started Succession
00:02:42.550 --> 00:02:45.969
together to really help bring a lot of this like.
00:02:46.270 --> 00:02:49.990
you know go -to -market tech tools uh training
00:02:49.990 --> 00:02:53.210
to the life science uh industry specifically
00:02:53.210 --> 00:02:55.810
so that's a lot of what we do today uh right
00:02:55.810 --> 00:02:58.870
now as we're talking we have a team of four and
00:02:58.870 --> 00:03:03.889
we have worked with around 75 different life
00:03:03.889 --> 00:03:06.650
science companies from everything from go to
00:03:06.650 --> 00:03:09.330
market so like running outbound campaigns paid
00:03:09.330 --> 00:03:13.259
ads content setting up different CRM systems
00:03:13.259 --> 00:03:16.699
and automations and sales training that we do
00:03:16.699 --> 00:03:20.500
as well. So far, I love it. It's amazing. And
00:03:20.500 --> 00:03:22.819
yeah, that's a bit. And I think one of the things
00:03:22.819 --> 00:03:25.479
we had talked about was when we started the agency,
00:03:25.699 --> 00:03:29.580
it was relatively early in the sort of AI hype
00:03:29.580 --> 00:03:32.939
cycle. And so I think Nick and I both realized
00:03:32.939 --> 00:03:35.879
very early on that we needed to be AI first.
00:03:36.400 --> 00:03:39.800
and ai forward so i know having your podcast
00:03:39.800 --> 00:03:42.460
in the early days as well was very useful so
00:03:42.460 --> 00:03:44.719
hearing you know you and martin's take on everything
00:03:44.719 --> 00:03:48.960
was you know i would say a great way for me to
00:03:48.960 --> 00:03:51.539
like understand all the different news and things
00:03:51.539 --> 00:03:53.039
that was going on and then i could go and dig
00:03:53.039 --> 00:03:55.439
deeper into it myself and then continue to to
00:03:55.439 --> 00:03:57.259
figure out how best we could use ai in our agency
00:03:57.259 --> 00:03:59.979
yeah i think one of the things i really loved
00:03:59.979 --> 00:04:02.659
about some of our
Load full transcript
early conversations that you
00:04:02.659 --> 00:04:06.860
know industry trade shows as an example is I
00:04:06.860 --> 00:04:09.719
would say you're a bit more like Martin than
00:04:09.719 --> 00:04:12.520
I am, which is that you two are very good at
00:04:12.520 --> 00:04:15.039
like having a vision for seeing how a process
00:04:15.039 --> 00:04:19.360
works now and being very good at iterating and
00:04:19.360 --> 00:04:22.259
testing different ways of bringing AI into that
00:04:22.259 --> 00:04:26.720
process. because you know every time i see you
00:04:26.720 --> 00:04:28.819
you've done some other really cool thing where
00:04:28.819 --> 00:04:31.000
i'm like i need to remember the details of this
00:04:31.000 --> 00:04:32.420
so i can talk about it on the podcast so it's
00:04:32.420 --> 00:04:34.120
much better to have you here to share the details
00:04:34.120 --> 00:04:36.240
so before we get into some of the ai bit i'm
00:04:36.240 --> 00:04:38.759
actually quite interested to know what from the
00:04:38.759 --> 00:04:42.120
sort of sales enablement operations rev ops type
00:04:42.120 --> 00:04:44.319
stuff what was the cool stuff that people were
00:04:44.319 --> 00:04:46.779
coming in and going hey this is cool that clearly
00:04:46.779 --> 00:04:49.319
most other people didn't have but you had So
00:04:49.319 --> 00:04:51.759
I would say even in the early days, it was something
00:04:51.759 --> 00:04:54.839
as simple as an email sequencing platform, something
00:04:54.839 --> 00:04:57.899
like a sales loft and outreach at the time were
00:04:57.899 --> 00:05:02.819
some of the main ones. And even today, we've
00:05:02.819 --> 00:05:05.040
trained companies that don't even have a tool
00:05:05.040 --> 00:05:07.300
to do that. And so there's still email sequencing
00:05:07.300 --> 00:05:10.399
platform is like the way where you can create
00:05:10.399 --> 00:05:13.439
a set of steps for a sequence. So like email
00:05:13.439 --> 00:05:16.139
one goes. goes out email two follows a couple
00:05:16.139 --> 00:05:18.180
days later email three follows after that right
00:05:18.180 --> 00:05:21.879
like a way to sort of organize um your top of
00:05:21.879 --> 00:05:24.360
funnel prospecting activity and we've talked
00:05:24.360 --> 00:05:26.480
to a couple companies many companies today actually
00:05:26.480 --> 00:05:28.100
that like don't even have these sorts of systems
00:05:28.100 --> 00:05:30.819
in place um typically they're more like the mid
00:05:30.819 --> 00:05:33.680
to large size um smaller companies i think are
00:05:33.680 --> 00:05:35.199
trying to figure out how to adopt this stuff
00:05:35.199 --> 00:05:38.100
quicker but it doesn't seem to be like the norm
00:05:38.100 --> 00:05:41.000
so like that was something really simple early
00:05:41.000 --> 00:05:44.769
on um I think another one was just how we sort
00:05:44.769 --> 00:05:50.069
of built out our training that we would do as
00:05:50.069 --> 00:05:53.170
well. So rather than having, say, a Google Doc
00:05:53.170 --> 00:05:55.490
with a bunch of links in it to go watch different
00:05:55.490 --> 00:05:57.509
trainings, or rather than just throwing somebody
00:05:57.509 --> 00:05:58.990
into the wolves, like, hey, now you're out in
00:05:58.990 --> 00:06:01.790
the field, go watch these people sell. We actually
00:06:01.790 --> 00:06:03.689
have structured training where we brought in
00:06:03.689 --> 00:06:06.689
different experts from the team, from across
00:06:06.689 --> 00:06:11.759
different departments as well. talk about the
00:06:11.759 --> 00:06:13.759
things that they were best at and train people
00:06:13.759 --> 00:06:16.459
at the things that they're best at and then structured
00:06:16.459 --> 00:06:19.480
it in like a formal like onboarding process that
00:06:19.480 --> 00:06:21.620
took place over a couple of months rather than
00:06:21.620 --> 00:06:23.720
like trying to fit it all within one week as
00:06:23.720 --> 00:06:26.379
an example. So I think those are some things
00:06:26.379 --> 00:06:28.459
off the top of my head. Yeah, that's really helpful
00:06:28.459 --> 00:06:30.379
because I think one of the things I really love
00:06:30.379 --> 00:06:31.939
to learn a bit more about today that I think
00:06:31.939 --> 00:06:34.480
would be valuable to the audience, even if most
00:06:34.480 --> 00:06:36.180
of our audience is working in marketing, I think
00:06:36.180 --> 00:06:37.980
a lot of these principles are going to be translatable.
00:06:38.259 --> 00:06:43.199
Where are you? now bringing AI into those, into
00:06:43.199 --> 00:06:45.120
the way you work, or some interesting processes
00:06:45.120 --> 00:06:48.019
and sort of tricks that you're using AI for.
00:06:48.740 --> 00:06:50.740
It's funny you actually talk about like, do these
00:06:50.740 --> 00:06:53.959
things apply to marketing? So I want to hit on
00:06:53.959 --> 00:06:59.360
that first, because I was probably like, I don't
00:06:59.360 --> 00:07:01.240
know, seven years ago, seven, eight years ago,
00:07:01.319 --> 00:07:04.339
it was like very much like the outbound motion
00:07:04.339 --> 00:07:07.970
should live in sales. Your SDR team should report
00:07:07.970 --> 00:07:10.509
into sales. And I was like, I was a hard stop.
00:07:10.670 --> 00:07:13.189
Should not report in marketing. That was my feeling
00:07:13.189 --> 00:07:19.850
at the time as a salesperson. However, now, I
00:07:19.850 --> 00:07:22.110
would say salespeople and sales leaders are typically
00:07:22.110 --> 00:07:26.029
not great with tech and tools and automation
00:07:26.029 --> 00:07:28.490
flows. It doesn't come naturally to a lot of
00:07:28.490 --> 00:07:31.459
them. Whereas marketing... a lot of marketing
00:07:31.459 --> 00:07:33.540
people are used to these things because they
00:07:33.540 --> 00:07:35.379
go into HubSpot, they've got to build out a workflow,
00:07:35.699 --> 00:07:38.560
email, nurture, whatever.
They're more, I think,
00:07:38.600 --> 00:07:42.899
tech forward. And so I think a lot of these AI
00:07:42.899 --> 00:07:46.319
flows and these systems for not only inbound
00:07:46.319 --> 00:07:50.420
follow -up, as an example, but also outbound
00:07:50.420 --> 00:07:53.180
should probably live under marketing. And you
00:07:53.180 --> 00:07:54.920
should probably figure out who is the person
00:07:54.920 --> 00:07:59.379
who's AI first, loves the systems. bit of a nerd
00:07:59.379 --> 00:08:01.980
when it comes to like, you know, more like engineering
00:08:01.980 --> 00:08:05.220
type of stuff, and put them in charge of building
00:08:05.220 --> 00:08:11.959
this stuff out. Because the possibilities are
00:08:11.959 --> 00:08:14.439
crazy, like you could literally probably put
00:08:14.439 --> 00:08:16.519
AI in any workflow that you could possibly imagine.
00:08:17.139 --> 00:08:19.480
And so I'm happy to talk through some that we
00:08:19.480 --> 00:08:22.600
come up with. But I think also, the way other
00:08:22.600 --> 00:08:25.300
people should be thinking about this is, what
00:08:25.300 --> 00:08:27.980
are the processes that you're doing today? that
00:08:27.980 --> 00:08:30.199
are like a pain in the butt and then the first
00:08:30.199 --> 00:08:33.659
step is okay let's go and plug in ai into certain
00:08:33.659 --> 00:08:36.120
areas of it let's use automation tools in certain
00:08:36.120 --> 00:08:38.460
areas of it to just like pick like two or three
00:08:38.460 --> 00:08:41.100
of them and just like go use it and do it and
00:08:41.100 --> 00:08:43.820
get something that works because i think what
00:08:43.820 --> 00:08:45.679
a lot of people will do is they're like overwhelmed
00:08:45.679 --> 00:08:48.440
with how much how the potential of how they could
00:08:48.440 --> 00:08:50.600
use ai there's a bajillion different tools every
00:08:50.600 --> 00:08:52.769
day there's a new one coming out like The only
00:08:52.769 --> 00:08:55.389
reason I have X on my phone is because that is
00:08:55.389 --> 00:08:57.429
where all of the AI news comes and everything
00:08:57.429 --> 00:09:00.190
hits there first, right? So like, yeah, I'm like
00:09:00.190 --> 00:09:02.490
a hype cycle guy. So like, I get all hyped up
00:09:02.490 --> 00:09:04.009
around new stuff. So like, I see the new stuff
00:09:04.009 --> 00:09:06.389
come and I'm like, gotta go check it out. But
00:09:06.389 --> 00:09:07.809
I think for the average person, it's probably
00:09:07.809 --> 00:09:10.690
not like great. And so pick a couple of workflows,
00:09:11.029 --> 00:09:13.570
automate them, get an idea of how you can start
00:09:13.570 --> 00:09:16.629
to leverage this stuff. And then like the second
00:09:16.629 --> 00:09:19.289
phase is now how can you start to do things that
00:09:19.289 --> 00:09:22.399
you couldn't previously do before? And I think
00:09:22.399 --> 00:09:24.779
that's sort of the areas that you can start to
00:09:24.779 --> 00:09:29.259
get into. I love that. So let's do a thought
00:09:29.259 --> 00:09:31.139
experiment. We're going to do it. We're going
00:09:31.139 --> 00:09:35.080
to do it live. We're going to pick a workflow
00:09:35.080 --> 00:09:36.799
and we're going to be like, that's a pain in
00:09:36.799 --> 00:09:39.779
the bum. Let's talk through where AI could help
00:09:39.779 --> 00:09:42.000
in it. And then I think we'll get creative and
00:09:42.000 --> 00:09:44.340
think about, well, what new capabilities can
00:09:44.340 --> 00:09:47.039
we do in that area with an AI? So are there any
00:09:47.039 --> 00:09:49.179
particular workflows that spring to mind where
00:09:49.179 --> 00:09:51.720
you're like, okay, cool, here's an example. Yeah.
00:09:51.799 --> 00:09:54.840
I mean, I think the perfect one is, I mean, the
00:09:54.840 --> 00:09:57.159
obvious one is like sales prospecting, but the
00:09:57.159 --> 00:09:58.779
way I think about this is probably a little bit
00:09:58.779 --> 00:10:02.299
different than maybe others. And so when you
00:10:02.299 --> 00:10:05.000
were, if you are a human SDR and you're going
00:10:05.000 --> 00:10:07.820
to go prospect to somebody. you're probably going
00:10:07.820 --> 00:10:10.299
to go do some amount of research on this person,
00:10:10.379 --> 00:10:11.919
right? So you're going to probably go to their
00:10:11.919 --> 00:10:13.620
company website. You'll probably go to their
00:10:13.620 --> 00:10:15.899
LinkedIn profile. You may go try and find different
00:10:15.899 --> 00:10:17.679
publications. If you have like a side of leads
00:10:17.679 --> 00:10:19.779
account or something like that, like where are
00:10:19.779 --> 00:10:21.720
you going to go to go find the research that
00:10:21.720 --> 00:10:24.220
you're going to do? My guess is it's probably
00:10:24.220 --> 00:10:26.820
in the range of 10 minutes of research that you're
00:10:26.820 --> 00:10:29.919
going to do on this person. Then you're going
00:10:29.919 --> 00:10:32.059
to think about how do I take that information
00:10:32.059 --> 00:10:36.279
and then write a email or a LinkedIn message
00:10:36.279 --> 00:10:38.440
or prepare yourself for a call, whatever it is,
00:10:38.460 --> 00:10:43.940
to that person. So today, AI can basically reverse
00:10:43.940 --> 00:10:48.460
engineer different automation tools, can reverse
00:10:48.460 --> 00:10:50.919
engineer that 10 minutes of research upfront,
00:10:51.220 --> 00:10:55.419
and then also draft that sort of context into
00:10:55.419 --> 00:10:59.019
a message for you to where you are editing versus
00:10:59.019 --> 00:11:03.299
doing all of that work yourself. So you can imagine
00:11:03.299 --> 00:11:04.740
you got like a hundred prospects you want to
00:11:04.740 --> 00:11:07.899
go after. You could have that list like ready
00:11:07.899 --> 00:11:10.440
for you to review in the morning rather than
00:11:10.440 --> 00:11:13.440
like having to spend whatever, a thousand minutes
00:11:13.440 --> 00:11:16.559
going through research on each of them. But what
00:11:16.559 --> 00:11:20.960
type of tools, like, is that, is that like a
00:11:20.960 --> 00:11:23.000
chat bot that you say, here's a list of people
00:11:23.000 --> 00:11:24.639
do the research and you get up in the morning
00:11:24.639 --> 00:11:26.259
and then you've got an output on or an email
00:11:26.259 --> 00:11:28.440
or, you know, how does that actually translate
00:11:28.440 --> 00:11:31.370
into real world application? Yeah, and I think
00:11:31.370 --> 00:11:34.309
this is also where a lot of people run into issues
00:11:34.309 --> 00:11:36.750
with AI is they'll try and shove all of this
00:11:36.750 --> 00:11:39.409
into a chat GPT where it's like, hey, here's
00:11:39.409 --> 00:11:42.629
the person I'm trying to reach out to. Do the
00:11:42.629 --> 00:11:45.029
research for me and write the message. And that
00:11:45.029 --> 00:11:47.950
is always going to be terrible. And I think that's
00:11:47.950 --> 00:11:50.289
where you get people that say, oh, it's so bad.
00:11:50.470 --> 00:11:52.549
AI messaging is terrible. And it's like, yeah,
00:11:52.669 --> 00:11:56.669
because you're using it terribly. If I think
00:11:56.669 --> 00:11:59.259
about... There's probably three core things that
00:11:59.259 --> 00:12:02.740
you have to have as just context for these models
00:12:02.740 --> 00:12:05.899
to work effectively. And if you think of any
00:12:05.899 --> 00:12:08.340
AI model, the context is the most important piece.
00:12:08.639 --> 00:12:10.700
That's all of the information that you're providing
00:12:10.700 --> 00:12:13.039
it upfront so that it can provide you with the
00:12:13.039 --> 00:12:16.059
output that you want. And I'd say those three
00:12:16.059 --> 00:12:18.759
things are information about your company and
00:12:18.759 --> 00:12:20.899
your product services, what you do, what you
00:12:20.899 --> 00:12:23.039
help solve, problems that you solve for people.
00:12:24.000 --> 00:12:25.600
information about the person that you're reaching
00:12:25.600 --> 00:12:27.419
out to right so that's going to be the company
00:12:27.419 --> 00:12:30.480
information their publications like i don't know
00:12:30.480 --> 00:12:32.559
different products they might use linkedin information
00:12:32.559 --> 00:12:35.519
that sort of thing and then how you wanted to
00:12:35.519 --> 00:12:38.639
write the message and i think people don't really
00:12:38.639 --> 00:12:40.940
think about these three things in tandem as being
00:12:40.940 --> 00:12:44.279
like a core part of like the context engine And
00:12:44.279 --> 00:12:46.279
the third one I was talking about is how you
00:12:46.279 --> 00:12:48.320
wanted to write the message is so important because
00:12:48.320 --> 00:12:50.519
AI is not going to be good at writing the message.
00:12:50.679 --> 00:12:53.720
So we have this 27 page document that we use
00:12:53.720 --> 00:12:57.080
to basically as contacts, we made a custom GPT
00:12:57.080 --> 00:13:00.860
with it to help us write the messages exactly
00:13:00.860 --> 00:13:03.879
how we want with different sets of frameworks
00:13:03.879 --> 00:13:05.519
and different ways that we might want to write
00:13:05.519 --> 00:13:07.860
it because not everybody responds to the exact
00:13:07.860 --> 00:13:09.879
same sets of messaging. So we need different
00:13:09.879 --> 00:13:12.019
variations of messaging that we're going to use
00:13:12.019 --> 00:13:14.759
in there. And again, that's just like one of
00:13:14.759 --> 00:13:16.639
the three pieces of context that I'm requiring.
00:13:17.379 --> 00:13:20.039
I love that. I'm glad that you mentioned like
00:13:20.039 --> 00:13:24.259
26, it's 27 page briefing document, basically.
00:13:24.399 --> 00:13:27.879
Like when you're saying you must give the models
00:13:27.879 --> 00:13:32.200
context, five bullet points instead of one bullet
00:13:32.200 --> 00:13:34.320
point is not what you're saying. You are saying
00:13:34.320 --> 00:13:39.259
basically an almost. Like a more thorough brief
00:13:39.259 --> 00:13:41.179
than you'd give a human, really, because, of
00:13:41.179 --> 00:13:42.919
course, humans have real world experience to
00:13:42.919 --> 00:13:44.679
draw upon. They might have written messages in
00:13:44.679 --> 00:13:46.419
the past that work for them that they can lean
00:13:46.419 --> 00:13:49.679
into. But actually, you're probably giving examples
00:13:49.679 --> 00:13:52.799
that it can mimic and all this other stuff all
00:13:52.799 --> 00:13:54.500
in this one document. Is that how that works?
00:13:55.159 --> 00:13:57.799
Yeah, there's essentially rules that we have
00:13:57.799 --> 00:14:00.940
of do's and don'ts. Then there are frameworks.
00:14:01.019 --> 00:14:03.659
Then there are example messages. And then there's
00:14:03.659 --> 00:14:07.059
also spam words to avoid. These are words that
00:14:07.059 --> 00:14:09.139
are going to get caught up in spam. If you use
00:14:09.139 --> 00:14:11.600
the word free in your message or you use the
00:14:11.600 --> 00:14:14.000
word promotion, those are more likely to end
00:14:14.000 --> 00:14:17.299
up in spam. all of these things sort of like
00:14:17.299 --> 00:14:20.519
cobbled together in this document um you're sort
00:14:20.519 --> 00:14:22.460
of like outline if you're not familiar with like
00:14:22.460 --> 00:14:26.080
markdown format um like markdown is a way that
00:14:26.080 --> 00:14:29.559
ai is really good at understanding how things
00:14:29.559 --> 00:14:31.580
are structured and so if you can structure your
00:14:31.580 --> 00:14:34.279
documents or your context in markdown format
00:14:34.279 --> 00:14:36.460
um it's going to be better too it's going to
00:14:36.460 --> 00:14:38.519
be easier for the ai to interpret understand
00:14:38.519 --> 00:14:42.220
that content i'm glad you mentioned markdown
00:14:42.220 --> 00:14:44.909
as well
i got two bookmarks that get hammered
00:14:44.909 --> 00:14:47.789
multiple times a day in my in my browser one
00:14:47.789 --> 00:14:50.870
turns google doc into markdown and the other
00:14:50.870 --> 00:14:53.490
one turns an entire web page into cleaned up
00:14:53.490 --> 00:14:58.129
markdown um so that you can give ai tools structured
00:14:58.129 --> 00:15:00.250
content that they can that they can understand
00:15:00.250 --> 00:15:02.490
and it's these i think these are the little things
00:15:02.490 --> 00:15:05.330
that you pick up over time that probably almost
00:15:05.330 --> 00:15:07.970
seem not meaningful to share with other people
00:15:07.970 --> 00:15:11.230
um but it's made a big difference for me yeah
00:15:12.450 --> 00:15:14.789
Yeah, we think of them as second nature, right?
00:15:14.850 --> 00:15:16.330
Because we've done it so much. So it's like,
00:15:16.389 --> 00:15:18.610
we've toyed around with them and played with
00:15:18.610 --> 00:15:20.610
them so much. And I think that's the other thing
00:15:20.610 --> 00:15:23.769
that a lot of people get stuck on is they try
00:15:23.769 --> 00:15:25.710
something doesn't look good, doesn't give them
00:15:25.710 --> 00:15:27.350
the output that they want. And they don't spend
00:15:27.350 --> 00:15:29.590
the time to really dig into these tools. Like,
00:15:29.649 --> 00:15:32.169
I posted something the other day, which I'm sure
00:15:32.169 --> 00:15:33.909
we can talk about here in a sec. But like, I
00:15:33.909 --> 00:15:36.149
think a very underutilized skill is like the
00:15:36.149 --> 00:15:39.090
technical prowess of understanding like APIs
00:15:39.090 --> 00:15:43.059
and webhooks. like how data flows between different
00:15:43.059 --> 00:15:46.860
systems because if you're able to do that it's
00:15:46.860 --> 00:15:49.240
a lot easier to like connect disparate data sources
00:15:49.240 --> 00:15:53.179
um and then also like pass data from one tool
00:15:53.179 --> 00:15:55.279
to another where one ai tool might be really
00:15:55.279 --> 00:15:57.100
good at one thing and another ai tool might be
00:15:57.100 --> 00:16:00.440
good at another thing um yeah so i think there's
00:16:00.440 --> 00:16:02.399
it's like an underrated skill there so are you
00:16:02.399 --> 00:16:04.299
building some of those types of things and can
00:16:04.299 --> 00:16:06.279
you give me an example i mean we i asked a minute
00:16:06.279 --> 00:16:10.129
ago about you know, having 100 leads and then
00:16:10.129 --> 00:16:13.149
coming back in the morning and having 100 prospecting
00:16:13.149 --> 00:16:15.289
emails to just double check, which you can't
00:16:15.289 --> 00:16:17.470
do with ChatGPT, right? It's a conversational
00:16:17.470 --> 00:16:19.590
interface, not a here's a big job, come back
00:16:19.590 --> 00:16:22.690
in a day with the results. But you could build
00:16:22.690 --> 00:16:25.309
something like that if you were comfortable using
00:16:25.309 --> 00:16:27.590
APIs and stuff. So are those the types of things
00:16:27.590 --> 00:16:31.490
you've been playing with? Yeah, I mean, our command
00:16:31.490 --> 00:16:34.830
center, if you will, is this tool called Clay.
00:16:35.320 --> 00:16:37.940
um where if you're in sales on linkedin you've
00:16:37.940 --> 00:16:40.220
probably at some point seen somebody post a clay
00:16:40.220 --> 00:16:43.320
table or talk about clay they've like gotten
00:16:43.320 --> 00:16:47.039
an insane valuation um over the last couple of
00:16:47.039 --> 00:16:49.460
months and years but essentially it's like a
00:16:49.460 --> 00:16:52.500
smart spreadsheet not to be confused with smart
00:16:52.500 --> 00:16:55.840
sheets but it's a smart spreadsheet where each
00:16:55.840 --> 00:16:58.559
row is like a record so it can either be a company
00:16:58.559 --> 00:17:01.289
or an individual person And then each column
00:17:01.289 --> 00:17:05.150
is a different data point or enrichment or where
00:17:05.150 --> 00:17:08.410
you can make a different API call. So as an example,
00:17:08.529 --> 00:17:10.970
let's add a LinkedIn profile for somebody I wanted
00:17:10.970 --> 00:17:13.690
to target. Well, I could use Clay to enrich that
00:17:13.690 --> 00:17:16.049
LinkedIn profile. Now I have all the information
00:17:16.049 --> 00:17:20.170
in Clay in this one place about them and their
00:17:20.170 --> 00:17:23.069
LinkedIn profile. I could then pull out, say,
00:17:23.089 --> 00:17:25.509
the company that they work at, then go and get
00:17:25.509 --> 00:17:27.509
the domain for that company, go to the company
00:17:27.509 --> 00:17:29.910
website, scrape the company website, understand
00:17:29.910 --> 00:17:31.930
what's going on, maybe go visit their pipeline
00:17:31.930 --> 00:17:34.450
page, what therapeutic areas are they going after,
00:17:34.569 --> 00:17:36.769
what targets are they going after. Those then
00:17:36.769 --> 00:17:38.990
get extrapolated into new fields within Clay.
00:17:40.349 --> 00:17:46.730
Then I could leverage different AI models to
00:17:46.730 --> 00:17:49.089
then interpret all of this information that we
00:17:49.089 --> 00:17:50.950
gathered about this person to come up with unique
00:17:50.950 --> 00:17:53.230
insights i could then feed it information about
00:17:53.230 --> 00:17:55.109
my company and you can sort of like build out
00:17:55.109 --> 00:17:57.970
this whole table where then at the end you have
00:17:57.970 --> 00:18:00.769
this like you know message that gets generated
00:18:00.769 --> 00:18:03.390
using all of this context from these different
00:18:03.390 --> 00:18:06.450
fields and then like whatever you could send
00:18:06.450 --> 00:18:09.309
that message to your graphs in gmail or you could
00:18:09.309 --> 00:18:11.069
send it straight into your email sequence or
00:18:11.069 --> 00:18:13.910
you could have it um like get pushed to a google
00:18:13.910 --> 00:18:16.109
doc that you could review
there's all sorts of
00:18:16.109 --> 00:18:18.269
like cool ways that you could do this stuff um
00:18:18.269 --> 00:18:21.490
and each of these other things along the way
00:18:21.490 --> 00:18:24.190
could also be a tool so like i know you're a
00:18:24.190 --> 00:18:27.269
big fan of heijin so heijin has an api so if
00:18:27.269 --> 00:18:30.329
you wanted to write a custom video message to
00:18:30.329 --> 00:18:33.730
somebody as an example using ai like heijin could
00:18:33.730 --> 00:18:36.210
be a you know one of these columns that you could
00:18:36.210 --> 00:18:38.670
enrich with a script to how you would send a
00:18:38.670 --> 00:18:41.589
message to that person and so By understanding
00:18:41.589 --> 00:18:43.529
how these tools talk to each other, you can do
00:18:43.529 --> 00:18:45.710
some really cool stuff. Yeah, that's awesome.
00:18:45.869 --> 00:18:49.170
So just to play that back, all these information
00:18:49.170 --> 00:18:52.490
sources being pulled into this sheet, it's interpreting
00:18:52.490 --> 00:18:55.490
them and then writing, say, a prospecting email.
00:18:55.609 --> 00:18:57.230
But like you say, you could even push that as
00:18:57.230 --> 00:18:58.849
a script. You could then write a script that
00:18:58.849 --> 00:19:00.569
then gets pushed to HeyGen, and then you've got
00:19:00.569 --> 00:19:03.789
a video embedded in your email as well. And if
00:19:03.789 --> 00:19:04.970
you want to write a different type of LinkedIn
00:19:04.970 --> 00:19:08.029
message, so you could have an example of, here's
00:19:08.029 --> 00:19:10.769
how we write emails. Here's how we write LinkedIn
00:19:10.769 --> 00:19:12.970
messages. Here's how we create video scripts.
00:19:13.210 --> 00:19:15.769
That could all be in one row. And then the output
00:19:15.769 --> 00:19:18.289
is now you've got three distinct messages across
00:19:18.289 --> 00:19:20.670
three different channels for that one individual.
00:19:21.509 --> 00:19:25.069
I love that. And I think that is a critical point
00:19:25.069 --> 00:19:30.089
because as these systems get easier to use, you
00:19:30.089 --> 00:19:31.750
know most people are going to be able to figure
00:19:31.750 --> 00:19:34.170
out a way of doing this as scale but probably
00:19:34.170 --> 00:19:36.849
averagely because they probably can't put that
00:19:36.849 --> 00:19:39.650
much time into the rules and the prompts whereas
00:19:39.650 --> 00:19:42.950
if you really want to be able to stand out it's
00:19:42.950 --> 00:19:46.910
the it's the 27 page context that drives the
00:19:46.910 --> 00:19:49.930
production of your script that makes it a world
00:19:49.930 --> 00:19:52.690
-class script that is the thing that really hits
00:19:52.690 --> 00:19:54.890
home with the audience not the fact that you
00:19:54.890 --> 00:19:57.940
can automate it as scale right Yeah, everybody's
00:19:57.940 --> 00:19:59.599
going to be able to do this stuff. They can already
00:19:59.599 --> 00:20:04.180
today, right? And I think if you get emails from
00:20:04.180 --> 00:20:07.819
people and you're like, this is clearly AI, that
00:20:07.819 --> 00:20:09.779
is, again, I think your point is that is the
00:20:09.779 --> 00:20:12.700
average, right? What you need to do is play with
00:20:12.700 --> 00:20:15.259
it so much and understand how these models work
00:20:15.259 --> 00:20:19.000
to such like a deep understanding of them where
00:20:19.000 --> 00:20:23.200
you can figure out how to force them to make
00:20:23.200 --> 00:20:25.859
it look human, right? It doesn't have those AI
00:20:25.859 --> 00:20:27.960
tendencies, right? So if you do this enough times,
00:20:27.980 --> 00:20:30.319
you can start to read all these AI tendencies.
00:20:30.460 --> 00:20:32.440
You see all these AI tendencies, and then you
00:20:32.440 --> 00:20:34.099
update your document, which is like, don't do
00:20:34.099 --> 00:20:36.099
this. Don't do that. Don't put an em dash, right?
00:20:36.359 --> 00:20:39.740
Don't say like, don't use this. Like, what is
00:20:39.740 --> 00:20:41.880
the cadence that it uses all the time? It loves
00:20:41.880 --> 00:20:45.240
to say, it's not just this. Yeah. Like that type
00:20:45.240 --> 00:20:47.140
of stuff. It's like, go human rights that way.
00:20:47.240 --> 00:20:49.779
Right. Or if it is, it's very few. And so you
00:20:49.779 --> 00:20:52.500
have to like. play with it and tweak it and constantly
00:20:52.500 --> 00:20:55.240
be updating this stuff. And I think people want
00:20:55.240 --> 00:20:58.759
AI to be the easy button, but it's not the easy
00:20:58.759 --> 00:21:01.920
button. It still requires a lot of work to constantly
00:21:01.920 --> 00:21:05.660
be iterating and improving it over time. Yeah,
00:21:05.720 --> 00:21:08.119
we used to call it, going back several years
00:21:08.119 --> 00:21:11.099
ago when Martin and I were doing training and
00:21:11.099 --> 00:21:13.460
consulting around this stuff, we used to call
00:21:13.460 --> 00:21:16.140
it the alien intern. And I think for all of the
00:21:16.140 --> 00:21:18.779
improvements, I still think it's an alien intern.
00:21:18.900 --> 00:21:24.640
It needs... context and guidance that the problem
00:21:24.640 --> 00:21:27.079
is you see things thrown around like gpt5 is
00:21:27.079 --> 00:21:29.960
at phd level and that has a bunch of connotations
00:21:29.960 --> 00:21:32.380
associated with it rightly or wrongly that it's
00:21:32.380 --> 00:21:34.660
going to have a certain intelligence or level
00:21:34.660 --> 00:21:37.900
of capabilities um it doesn't have 15 years of
00:21:37.900 --> 00:21:40.400
sales prospecting experience it doesn't have
00:21:40.400 --> 00:21:42.900
15 years of marketing experience in the real
00:21:42.900 --> 00:21:46.160
world seeing what works and what doesn't what
00:21:46.160 --> 00:21:47.980
it does have though is it has the average of
00:21:47.980 --> 00:21:51.140
all of this stuff right which is i
think where
00:21:51.140 --> 00:21:54.740
you get the average of the result so it's like
00:21:54.740 --> 00:21:58.160
it's got phd level experience meaning like it
00:21:58.160 --> 00:22:00.559
can answer really complex problems and questions
00:22:00.559 --> 00:22:02.619
but then everything else is trained on is the
00:22:02.619 --> 00:22:05.019
average so it's like it's been trained on millions
00:22:05.019 --> 00:22:07.839
probably of average emails not like the best
00:22:07.839 --> 00:22:10.559
emails or it's been trained on millions of like
00:22:10.559 --> 00:22:12.720
marketing campaigns that are like the average
00:22:12.720 --> 00:22:15.980
marketing campaign and so like to your point
00:22:16.480 --> 00:22:18.380
how do you get how do you be excellent it's like
00:22:18.380 --> 00:22:20.440
it still has that human component how do you
00:22:20.440 --> 00:22:25.059
how do you use the human creativity to like then
00:22:25.059 --> 00:22:27.079
leverage ai with that like you still need the
00:22:27.079 --> 00:22:29.240
human creativity element yeah it's going to be
00:22:29.240 --> 00:22:30.920
like a normal distribution curve probably right
00:22:30.920 --> 00:22:32.900
it's like it's probably got access to loads of
00:22:32.900 --> 00:22:35.079
blog posts about how to write the best emails
00:22:35.079 --> 00:22:39.579
um but by definition you know the very best emails
00:22:39.579 --> 00:22:42.579
are the small amount of data in that population
00:22:42.579 --> 00:22:44.539
right at the far end of that normal distribution
00:22:45.279 --> 00:22:46.859
And when it's trying to work out the patterns,
00:22:47.019 --> 00:22:49.259
it's going to drift into the middle. And that's
00:22:49.259 --> 00:22:52.319
what's going to give you middling results. We
00:22:52.319 --> 00:22:55.940
touched on this in a few ways there, but I'd
00:22:55.940 --> 00:23:00.279
love to get your take on the importance of humans
00:23:00.279 --> 00:23:05.119
and creativity. And what constitutes the misuse
00:23:05.119 --> 00:23:08.819
of AI in sales prospecting and sales process?
00:23:09.549 --> 00:23:12.650
Copy and pasting and sending is the misuse, right?
00:23:12.829 --> 00:23:15.309
It's like I go, I get the result. I'm going to
00:23:15.309 --> 00:23:17.289
copy it. I'm going to paste it and I hit send,
00:23:17.450 --> 00:23:21.210
or I just assume that it's correct. AI still
00:23:21.210 --> 00:23:23.150
does hallucinate no matter how much content and
00:23:23.150 --> 00:23:24.509
information that you give it. So you'd still
00:23:24.509 --> 00:23:27.849
have to double check these things. And so I think
00:23:27.849 --> 00:23:31.230
that is probably the biggest misuse. That's probably
00:23:31.230 --> 00:23:33.819
across the board too. It's like I've seen. stuff
00:23:33.819 --> 00:23:36.960
is just like factually inaccurate um that's like
00:23:36.960 --> 00:23:39.119
posted online and blog posts and stuff like well
00:23:39.119 --> 00:23:41.240
i don't
know where you got that from that's not
00:23:41.240 --> 00:23:44.019
real um and so i think there's still like one
00:23:44.019 --> 00:23:47.180
the human in the loop to verify a lot of things
00:23:47.180 --> 00:23:51.000
but then also like it is not that creative right
00:23:51.000 --> 00:23:53.059
it's trained on stuff that's already occurred
00:23:53.059 --> 00:23:56.400
so it's not going to come up with new novel concepts
00:23:56.400 --> 00:23:58.460
very well maybe that'll change in the future
00:23:58.460 --> 00:24:00.180
but i think today it's just not really there
00:24:00.180 --> 00:24:03.619
and so you still need the human creativity which
00:24:03.619 --> 00:24:06.460
I think comes from really understanding your
00:24:06.460 --> 00:24:09.420
customer and what your customer wants and needs
00:24:09.420 --> 00:24:11.720
and how your customer thinks, and then figuring
00:24:11.720 --> 00:24:15.660
out how you can provide novel, interesting content,
00:24:15.960 --> 00:24:21.880
insights, value to them. And then you basically
00:24:21.880 --> 00:24:25.140
force AI to your will, if you will, right? Which
00:24:25.140 --> 00:24:26.960
is like, how do I force it to get the outputs
00:24:26.960 --> 00:24:29.180
that I want that achieve that outcome versus
00:24:29.180 --> 00:24:31.240
going to it and like, give me five ideas on a
00:24:31.240 --> 00:24:35.349
blog post idea, you know? Yeah, in a slightly
00:24:35.349 --> 00:24:37.130
tangential way, I've been watching some of the
00:24:37.130 --> 00:24:39.970
Sora 2 examples that have been coming out. One
00:24:39.970 --> 00:24:45.130
of my favorites is, it's like a cops style sort
00:24:45.130 --> 00:24:50.069
of road chase with Super Mario. And they try
00:24:50.069 --> 00:24:52.369
to pull him over. And it's Super Mario from Mario
00:24:52.369 --> 00:24:54.089
Kart. And if you haven't seen it, you should
00:24:54.089 --> 00:24:56.750
Google it. It's absolutely brilliant. Why do
00:24:56.750 --> 00:24:59.829
I like that? If AI had come up with that idea
00:24:59.829 --> 00:25:02.470
and then executed it. it would mean far less
00:25:02.470 --> 00:25:06.589
to me than me thinking, who is the genius who
00:25:06.589 --> 00:25:08.829
thought, you know what would be cool? I'll ask
00:25:08.829 --> 00:25:12.490
Sora to bring this idea to life for me. That's
00:25:12.490 --> 00:25:15.170
the interesting part to me. I don't want to go
00:25:15.170 --> 00:25:18.569
and watch a load of AI videos by AI, even if
00:25:18.569 --> 00:25:22.029
AI gets quite creative, because it's like the
00:25:22.029 --> 00:25:24.829
respect that I have for the creativity of the
00:25:24.829 --> 00:25:27.269
person who leveraged the tool that makes me like
00:25:27.269 --> 00:25:31.369
it, not just AI on. on steroids doing everything
00:25:31.369 --> 00:25:33.789
on its own and to your point i think when you
00:25:33.789 --> 00:25:38.549
get an email from a from a from an sdr or whoever
00:25:38.549 --> 00:25:41.430
that's been well thought out well researched
00:25:41.430 --> 00:25:45.430
if it tries to be clever or funny really feels
00:25:45.430 --> 00:25:48.509
genuinely human in it then you can empathize
00:25:48.509 --> 00:25:50.329
with it if you think the ai just came up with
00:25:50.329 --> 00:25:53.029
the jokes this doesn't hits the same right yeah
00:25:53.029 --> 00:25:56.150
jokes suck they're really bad it's very rare
00:25:56.150 --> 00:25:59.880
that you find a good one absolutely i think to
00:25:59.880 --> 00:26:02.359
your point like uh along the lines of the mario
00:26:02.359 --> 00:26:04.680
video uh utah i think you guys talked about this
00:26:04.680 --> 00:26:07.720
in your last episode which is the the yeti um
00:26:07.720 --> 00:26:10.799
like there was i can't remember the brand now
00:26:10.799 --> 00:26:13.099
which is strange but i remember the commercial
00:26:13.099 --> 00:26:15.220
they made like dozens of these different commercials
00:26:15.220 --> 00:26:19.339
with this like giant yeti you know snow abominable
00:26:19.339 --> 00:26:22.559
snowmen thing and uh like it was very clearly
00:26:22.559 --> 00:26:24.759
ai but it was so creative and interesting it's
00:26:24.759 --> 00:26:27.900
like him skydiving out of a plane and like holding
00:26:27.900 --> 00:26:30.119
up like the product or whatever right like there's
00:26:30.119 --> 00:26:32.799
some there's really cool like brand stuff that
00:26:32.799 --> 00:26:35.299
they were able to do which was clearly ai but
00:26:35.299 --> 00:26:37.119
it was creative and so then it was able to go
00:26:37.119 --> 00:26:39.839
viral so you kind of have this concept like this
00:26:39.839 --> 00:26:41.680
we're probably at this interesting inflection
00:26:41.680 --> 00:26:44.599
point where you could really get creative with
00:26:44.599 --> 00:26:48.259
like how your team is using ai to capture attention
00:26:49.339 --> 00:26:51.960
And as you probably know, in the life sciences,
00:26:51.980 --> 00:26:55.660
like the general marketing is quite bland a lot
00:26:55.660 --> 00:26:58.579
of the time. Right. A lot of people aren't willing
00:26:58.579 --> 00:27:03.339
to push the needle on the creative side and be
00:27:03.339 --> 00:27:07.299
risky. What a massive opportunity. Plus, I also
00:27:07.299 --> 00:27:09.460
think there is appetite for people who want to
00:27:09.460 --> 00:27:12.700
see something interesting, even from life science
00:27:12.700 --> 00:27:14.980
brands. I think one of the common pushbacks you'll
00:27:14.980 --> 00:27:17.200
always get is, you know, scientists are not.
00:27:17.450 --> 00:27:19.589
looking for us to entertain them they want the
00:27:19.589 --> 00:27:22.690
most robust reliable reagents and instruments
00:27:22.690 --> 00:27:24.650
and tools so that they can trust their research
00:27:24.650 --> 00:27:27.109
and it's like why can't it be both why is it
00:27:27.109 --> 00:27:31.009
why does it have to be one right yeah they still
00:27:31.009 --> 00:27:33.569
scroll tiktok and instagram and whatever they
00:27:33.569 --> 00:27:35.930
laugh at beams like same thing for executives
00:27:35.930 --> 00:27:37.690
like people think you have to be so stuffy when
00:27:37.690 --> 00:27:39.710
you're reaching out to executives again they
00:27:39.710 --> 00:27:42.069
are people they laugh at jokes like they have
00:27:42.069 --> 00:27:45.079
senses of humor I don't think that everything
00:27:45.079 --> 00:27:51.900
has to be stuffy, perfectly rehearsed and pass
00:27:51.900 --> 00:27:54.819
through 17 rounds of legal review before it's
00:27:54.819 --> 00:27:57.960
able to see the light of day. There's always
00:27:57.960 --> 00:27:59.720
going to be businesses that have to do that.
00:28:00.240 --> 00:28:02.380
And there are always going to be businesses that,
00:28:02.440 --> 00:28:04.359
for whatever reason, absolutely feel they need
00:28:04.359 --> 00:28:06.839
to err on the side of caution. But if you're
00:28:06.839 --> 00:28:09.140
in a small or mid -sized business and you don't
00:28:09.140 --> 00:28:11.250
necessarily have those constraints, this has
00:28:11.250 --> 00:28:13.369
to be a competitive advantage that you at least
00:28:13.369 --> 00:28:17.990
look into, right? Because we're so deluged with
00:28:17.990 --> 00:28:19.769
content and that's going to get worse and worse
00:28:19.769 --> 00:28:22.769
and worse with AI. You have to break the pattern.
00:28:23.390 --> 00:28:25.710
And one way to do that is to connect with people
00:28:25.710 --> 00:28:27.609
emotionally or to make them laugh, right? And
00:28:27.609 --> 00:28:30.450
so ultimately, as much as the feature benefit
00:28:30.450 --> 00:28:32.750
set of your latest microscope is probably really
00:28:32.750 --> 00:28:35.769
interesting for the research applications it
00:28:35.769 --> 00:28:38.690
opens up and enables, that's not pattern matching
00:28:38.690 --> 00:28:41.579
in amongst all this other stuff, right? yeah
00:28:41.579 --> 00:28:43.480
it's like if you're to your point if you're going
00:28:43.480 --> 00:28:47.579
up against the you know big companies the thermos
00:28:47.579 --> 00:28:50.359
whoever the world like and you know that they
00:28:50.359 --> 00:28:52.539
have to have their like you know primmed and
00:28:52.539 --> 00:28:55.200
proper messaging that goes out and their specific
00:28:55.200 --> 00:28:58.579
ad copy that they have to use do something different
00:28:58.579 --> 00:29:01.099
right if you just go with the same thing using
00:29:01.099 --> 00:29:03.099
your color scheme like it's not going to stand
00:29:03.099 --> 00:29:06.980
out for sure um so just coming back quickly then
00:29:06.980 --> 00:29:09.559
to the misuse of ai because we talked about hygiene
00:29:09.559 --> 00:29:14.289
earlier and synthetic rep videos um you know
00:29:14.289 --> 00:29:16.630
a video that looks like harrison speaks like
00:29:16.630 --> 00:29:20.230
harrison is not harrison how do
you feel about
00:29:20.230 --> 00:29:23.990
the use of the current set of video capabilities
00:29:23.990 --> 00:29:25.670
and then where do you think this is going to
00:29:25.670 --> 00:29:28.869
go in the future interesting i think that it's
00:29:28.869 --> 00:29:32.809
generally pretty good i think that unless somebody
00:29:32.809 --> 00:29:35.569
knows you it's hard to pick up some of the nuances
00:29:35.569 --> 00:29:37.750
like you will watch the video and be like that's
00:29:37.750 --> 00:29:39.569
not me like i don't talk like that that's not
00:29:39.569 --> 00:29:41.670
quite who i am but if you're sending it to somebody
00:29:41.670 --> 00:29:43.789
who has no idea who you are they don't know like
00:29:43.789 --> 00:29:48.190
your nuances or how your inflection in your voice
00:29:48.190 --> 00:29:52.049
um shows up and and it's only going to get better
00:29:52.049 --> 00:29:55.650
at that too but then The other thing I think
00:29:55.650 --> 00:29:58.769
you can lean into is just clearly saying, this
00:29:58.769 --> 00:30:02.349
is AI. I made this specifically with AI for you.
00:30:02.609 --> 00:30:04.930
Let me know what you think. People like something
00:30:04.930 --> 00:30:07.490
novel or something interesting. Just lean into
00:30:07.490 --> 00:30:10.670
it and add another piece of flavor when you're
00:30:10.670 --> 00:30:14.609
sending it in the text that calls out that it's
00:30:14.609 --> 00:30:17.269
AI or something. But then add something personal
00:30:17.269 --> 00:30:19.470
yourself in that text or something. There's a
00:30:19.470 --> 00:30:24.950
way that you can do this that stands out. And
00:30:24.950 --> 00:30:27.789
like you call it, like if you're like in a feed,
00:30:27.849 --> 00:30:29.690
right? It's like, I'll stop the scroll. It's
00:30:29.690 --> 00:30:31.269
like, I'm scrolling and it stops the scroll.
00:30:31.589 --> 00:30:33.769
Same thing. If I get an email and it's like,
00:30:33.769 --> 00:30:35.829
looks different. I'm going to like click on it
00:30:35.829 --> 00:30:37.730
and see what's up. Same thing on LinkedIn. If
00:30:37.730 --> 00:30:39.690
I'm in my messages and it looks different, I
00:30:39.690 --> 00:30:42.390
might click on it and see what it's about. So
00:30:42.390 --> 00:30:44.230
I think if you can find novel ways to do that,
00:30:44.289 --> 00:30:47.369
it's just going to improve your chances. Yeah.
00:30:47.430 --> 00:30:49.130
I think it creates quite an interesting arms
00:30:49.130 --> 00:30:52.170
race. I remember the first time I got a. like
00:30:52.170 --> 00:30:55.890
a wistier video that was just for me. And I watched
00:30:55.890 --> 00:30:58.190
it and I engaged with that company. They had
00:30:58.190 --> 00:30:59.750
a fairly interesting value. They've done the
00:30:59.750 --> 00:31:01.589
research on Biostrata. And I thought, actually,
00:31:01.650 --> 00:31:03.450
yeah, this person's put the effort in here. I'm
00:31:03.450 --> 00:31:05.609
going to engage in that. But it doesn't take
00:31:05.609 --> 00:31:07.450
very long for that to be overused and then you
00:31:07.450 --> 00:31:10.470
ignore that, right? One of the things I've toyed
00:31:10.470 --> 00:31:12.509
with, but I think I never did it because it just
00:31:12.509 --> 00:31:15.029
felt too risky, was like using Suno and creating
00:31:15.029 --> 00:31:17.769
a song about the person that I'm targeting and
00:31:17.769 --> 00:31:19.589
their brand and how we could work together for
00:31:19.589 --> 00:31:22.759
an amazing future. having it like a quite a funny
00:31:22.759 --> 00:31:25.400
like upbeat country or something like that um
00:31:25.400 --> 00:31:28.599
but then but why not but why not why wouldn't
00:31:28.599 --> 00:31:30.980
you do it i'm just curious why wouldn't i do
00:31:30.980 --> 00:31:33.460
it i don't know i didn't do a huge amount of
00:31:33.460 --> 00:31:35.180
prospecting anyway very lucky i had a lot of
00:31:35.180 --> 00:31:39.920
inbound traffic um and it was just about risk
00:31:39.920 --> 00:31:42.660
management for me it's like risk reward i didn't
00:31:42.660 --> 00:31:45.160
really need i mostly thought it was a cool idea
00:31:45.160 --> 00:31:48.740
rather than like a deep need in me like we need
00:31:48.740 --> 00:31:50.779
the business i've got to i've got to break through
00:31:50.779 --> 00:31:53.759
the noise this is how i'm going to do it challenge
00:31:53.759 --> 00:31:56.619
a little bit like what's the downside they respond
00:31:56.619 --> 00:32:00.720
to be like wow this song sucks like they're probably
00:32:00.720 --> 00:32:03.119
not gonna i mean not gonna hold it against you
00:32:03.119 --> 00:32:05.079
as a company probably right and i think if you
00:32:05.079 --> 00:32:08.099
own it like hey you know we'd love to work together
00:32:08.099 --> 00:32:11.319
and i made this like really fun little ai song
00:32:11.319 --> 00:32:13.480
for you you might hate it but let me know what
00:32:13.480 --> 00:32:16.950
you think like that kind of message is uh I think
00:32:16.950 --> 00:32:20.109
no downside in that. And if the person honestly
00:32:20.109 --> 00:32:24.069
doesn't respond super negatively to it, then
00:32:24.069 --> 00:32:26.849
they suck as a person and you probably don't
00:32:26.849 --> 00:32:29.869
want to work with them anyway. Now I want you
00:32:29.869 --> 00:32:32.230
to go test this theory for me and then maybe
00:32:32.230 --> 00:32:33.869
it works really well and then it gets saturated
00:32:33.869 --> 00:32:36.509
and then you and no one else can do it anymore.
00:32:36.650 --> 00:32:38.250
And that's where the arms race comes in, I think.
00:32:38.549 --> 00:32:40.829
Yeah, but I would challenge that too and say
00:32:40.829 --> 00:32:43.680
like... Video messaging has been around for so
00:32:43.680 --> 00:32:47.759
long and nobody gets them. Nobody uses it. It
00:32:47.759 --> 00:32:52.359
is still like Greenfield, Blue Ocean. Nobody
00:32:52.359
--> 00:32:54.740
does it. And it's like one of the best ways that
00:32:54.740 --> 00:32:58.700
you can be personable to somebody. I think that
00:32:58.700 --> 00:33:02.049
you and I... live in a bubble of like we see
00:33:02.049 --> 00:33:03.849
this stuff and we're like probably people are
00:33:03.849 --> 00:33:06.250
going to do this and so if they do it if we do
00:33:06.250 --> 00:33:08.250
it then everybody's going to do it but nobody
00:33:08.250 --> 00:33:10.210
does it i don't think people catch on to this
00:33:10.210 --> 00:33:15.009
stuff i think that um it's very rare that if
00:33:15.009 --> 00:33:17.109
you come up with something kind of really interesting
00:33:17.109 --> 00:33:19.390
and novel to your point of sending like a song
00:33:19.390 --> 00:33:21.910
that's custom made for somebody i don't think
00:33:21.910 --> 00:33:25.930
very many people will do it and even if 10 people
00:33:25.930 --> 00:33:28.900
do it or 20 people do it It's not going to oversaturate,
00:33:28.900 --> 00:33:31.700
you know. I guess you're right. Why do you think
00:33:31.700 --> 00:33:34.779
that video and these other things that to you
00:33:34.779 --> 00:33:36.700
and I seem like they must be saturated by now,
00:33:36.759 --> 00:33:39.559
why are people not doing it? It takes work. You
00:33:39.559 --> 00:33:40.819
think that's when it comes down to it? That's
00:33:40.819 --> 00:33:43.460
the answer. Yeah, most people are lazy. I think
00:33:43.460 --> 00:33:46.220
that's the answer. Most people are lazy and they
00:33:46.220 --> 00:33:47.799
just won't put in the effort to actually do it.
00:33:48.430 --> 00:33:50.789
So I think it's really easy. It's really easy
00:33:50.789 --> 00:33:53.029
to have AI write text and load up an email and
00:33:53.029 --> 00:33:55.549
send it. It's a lot harder to connect with somebody
00:33:55.549 --> 00:33:58.950
on LinkedIn and then click this button that makes
00:33:58.950 --> 00:34:01.190
you feel very vulnerable that pops a screen up
00:34:01.190 --> 00:34:03.890
in front of you where you now have to like create
00:34:03.890 --> 00:34:06.450
a video and then send it to that person. So we
00:34:06.450 --> 00:34:08.269
can, in theory, we're going to get to test your
00:34:08.269 --> 00:34:10.550
hypothesis because if all of these tools made
00:34:10.550 --> 00:34:13.730
this almost click button easy, right? Like here's
00:34:13.730 --> 00:34:15.809
a prospect research, basically your whole clay
00:34:15.809 --> 00:34:18.679
work for you mentioned. but also right through
00:34:18.679 --> 00:34:21.539
to recording a video and in like an ai video
00:34:21.539 --> 00:34:25.039
and sending it um that should be click of the
00:34:25.039 --> 00:34:26.679
button and then we'll see if people will actually
00:34:26.679 --> 00:34:31.559
engage and do it again i don't know people are
00:34:31.559 --> 00:34:33.320
going to take the time to do it but yeah if they
00:34:33.320 --> 00:34:36.530
do great and if you do
this send us like send
00:34:36.530 --> 00:34:38.369
us the results i'd love to see it but i just
00:34:38.369 --> 00:34:40.610
i still don't think enough people are going to
00:34:40.610 --> 00:34:42.409
spend the time to learn how to set this up in
00:34:42.409 --> 00:34:44.309
clay learn how to connect to different systems
00:34:44.309 --> 00:34:47.150
and so if you are one of those people you have
00:34:47.150 --> 00:34:50.610
a very distinct advantage and this is why i keep
00:34:50.610 --> 00:34:52.889
talking about on a regular basis is like this
00:34:52.889 --> 00:34:55.789
is an opportunity to completely revamp your skill
00:34:55.789 --> 00:34:59.869
set that has probably never happened in my lifetime
00:34:59.869 --> 00:35:02.230
other than maybe like the start of the internet
00:35:02.230 --> 00:35:04.590
right it's like like this hasn't really happened
00:35:04.590 --> 00:35:08.789
and so Yeah, I think it's a really interesting
00:35:08.789 --> 00:35:11.409
opportunity for people to rebuild their skillset,
00:35:11.489 --> 00:35:13.789
to really stand out. And then you can leverage
00:35:13.789 --> 00:35:17.030
that to get like pretty much any job you want
00:35:17.030 --> 00:35:20.329
if you become really good at it. So here we go
00:35:20.329 --> 00:35:24.289
then. Obviously, like everyone, Harris and I
00:35:24.289 --> 00:35:26.849
get a lot of LinkedIn messages and a lot of emails
00:35:26.849 --> 00:35:30.070
of people trying to sell us stuff. We will reply
00:35:30.070 --> 00:35:34.670
to you if you send us a song about us. I'm in.
00:35:35.019 --> 00:35:37.519
Dude, I would love that. Honestly, one way that
00:35:37.519 --> 00:35:38.760
people could do it, I'll tell you, here's the
00:35:38.760 --> 00:35:42.199
playbook. Go take this. This will get posted,
00:35:42.300 --> 00:35:44.820
I think, on YouTube, right? Yeah. Yeah, this
00:35:44.820 --> 00:35:47.980
will get on YouTube. So go and get the transcript
00:35:47.980 --> 00:35:52.059
for this video from YouTube. Then take that transcript,
00:35:52.340 --> 00:35:55.920
put it into any of these. What is it? Suno, I
00:35:55.920 --> 00:35:58.389
think, is the one, right? yeah you know how to
00:35:58.389 --> 00:36:00.349
write a have a write a script have it write a
00:36:00.349 --> 00:36:02.130
song or put it in chat you have to write the
00:36:02.130 --> 00:36:06.050
song then upload that into suno then in suno
00:36:06.050 --> 00:36:08.150
have it create the song and then send us the
00:36:08.150 --> 00:36:11.429
song about this podcast that would be that that's
00:36:11.429 --> 00:36:13.730
the that's the challenge for people yeah that
00:36:13.730 --> 00:36:15.389
would be cool all right i expect to see that
00:36:15.389 --> 00:36:19.460
now And if Suno has an API, again, you could
00:36:19.460 --> 00:36:22.480
do this in Clay for all of your prospects. In
00:36:22.480 --> 00:36:24.219
Clay, you'd just basically be able to create
00:36:24.219 --> 00:36:26.519
these custom songs. I actually kind of do this,
00:36:26.539 --> 00:36:28.280
so I might start sending some songs to people.
00:36:29.079 --> 00:36:33.099
I want to know how well that works, especially
00:36:33.099 --> 00:36:35.219
because maybe the arms race is not as real as
00:36:35.219 --> 00:36:37.579
I thought it was, and we don't have to stay ahead
00:36:37.579 --> 00:36:40.579
of people's attention and BS filters quite at
00:36:40.579 --> 00:36:45.539
the level I'd imagined. Another thing I was thinking
00:36:45.539 --> 00:36:48.219
when I was talking about clay, maybe we have
00:36:48.219 --> 00:36:50.199
to go and do this from Rafael Harrison, it can
00:36:50.199 --> 00:36:53.019
be our weekend project, is whether or not you
00:36:53.019 --> 00:36:56.139
could like vibe code something up that would
00:36:56.139 --> 00:36:58.760
achieve this type of outcome, right? Well, okay,
00:36:58.840 --> 00:37:00.460
wait, where are you going with this? Vibe coding
00:37:00.460 --> 00:37:03.480
the front end of like our idea here or vibe coding
00:37:03.480 --> 00:37:08.159
something from clay? I've meant the entire front
00:37:08.159 --> 00:37:10.949
and back end. where it probably uses like an
00:37:10.949 --> 00:37:14.250
air table database or whatever but can you vibe
00:37:14.250 --> 00:37:16.570
code and plug straight into clay because then
00:37:16.570 --> 00:37:20.769
it would be even easier right two okay well i'll
00:37:20.769 --> 00:37:22.769
take this in two ways because i've seen something
00:37:22.769 --> 00:37:24.809
really cool i've never done it because it doesn't
00:37:24.809 --> 00:37:27.190
really apply to me i don't think or i couldn't
00:37:27.190 --> 00:37:31.380
figure out how to apply it as of yet This guy,
00:37:31.440 --> 00:37:34.400
if you're familiar with Lovable, it is one of
00:37:34.400 --> 00:37:36.139
these vibe coding platforms where basically you
00:37:36.139 --> 00:37:38.260
can program in natural language. You type in
00:37:38.260 --> 00:37:39.760
what you wanted to create, it will go create
00:37:39.760 --> 00:37:43.360
that application for you. So he used the Lovable
00:37:43.360 --> 00:37:47.840
API, used it in clay. So he took a company and
00:37:47.840 --> 00:37:50.719
then basically created a prop, like basically
00:37:50.719 --> 00:37:53.260
understood what is it that this company or a
00:37:53.260 --> 00:37:55.079
tool that would be really useful for this company.
00:37:56.079 --> 00:37:59.059
That created whatever, like the idea for the
00:37:59.059 --> 00:38:01.760
tool. Then that tool creates like, okay, there'll
00:38:01.760 --> 00:38:04.320
be a prompt for how I could build this tool that
00:38:04.320 --> 00:38:06.360
could be used in a vibe coding platform that
00:38:06.360 --> 00:38:09.739
generated the prompt. Then the prompt was sent
00:38:09.739 --> 00:38:12.500
to lovable to go and build it. And then it returned
00:38:12.500 --> 00:38:17.429
the deployed URL of that app. So literally. you
00:38:17.429 --> 00:38:20.670
can make custom apps for prospects. So like,
00:38:20.690 --> 00:38:22.469
if you're like a software development company
00:38:22.469 --> 00:38:25.250
or something, like imagine how, and I wanted
00:38:25.250 --> 00:38:28.449
to prospect into like 50 companies, I get a list
00:38:28.449 --> 00:38:30.409
of these 50 companies. I can go and just like
00:38:30.409 --> 00:38:34.010
create 50 MVPs of tools that are useful for them.
00:38:34.309 --> 00:38:35.690
And like, you're probably not going to send it
00:38:35.690 --> 00:38:37.449
right away from there. Like you go in and use
00:38:37.449 --> 00:38:39.289
it. And I just add a couple like tweaks here
00:38:39.289 --> 00:38:41.070
or there. And I've got something that I can send
00:38:41.070 --> 00:38:44.610
as like a really cool lead magnet for them. That's
00:38:44.610 --> 00:38:47.289
a cool idea. I think that's why people who really
00:38:47.289 --> 00:38:51.250
think creatively and come up with these things,
00:38:51.369 --> 00:38:53.349
they don't think, because the other thing is
00:38:53.349 --> 00:38:56.289
like, you could build that and maybe it breaks
00:38:56.289 --> 00:39:00.610
a lot, but in a year it won't. And this is how
00:39:00.610 --> 00:39:02.130
you have to think about AI, right? If you're
00:39:02.130 --> 00:39:04.510
constantly pushing at the capabilities of what
00:39:04.510 --> 00:39:06.809
these different tools, Lovable, Chachi, BT, Sora
00:39:06.809 --> 00:39:09.929
can do, but it's a bit clunky and broken, doesn't
00:39:09.929 --> 00:39:12.260
quite give you the output that you want. You
00:39:12.260 --> 00:39:13.820
have just built something that you'll be able
00:39:13.820 --> 00:39:15.840
to deploy in six months instead when you can
00:39:15.840 --> 00:39:19.019
swap the latest models in. Yeah, like if you
00:39:19.019 --> 00:39:21.179
had built on the original HeyGen, probably wouldn't
00:39:21.179 --> 00:39:23.019
have been very good to send to people. But now
00:39:23.019 --> 00:39:25.059
it's pretty darn good and now it integrates with
00:39:25.059 --> 00:39:27.760
SOAR too. So like you could leverage it for that
00:39:27.760 --> 00:39:30.820
even. So yeah, wild. Yeah, that would be cool.
00:39:31.440 --> 00:39:33.360
The other thing that you talked about was actually
00:39:33.360 --> 00:39:38.239
vibe coding in application. So the way that,
00:39:38.280 --> 00:39:41.940
so the short answer is yes. I am not an engineer
00:39:41.940 --> 00:39:46.280
at all. However, I have built probably like,
00:39:46.280 --> 00:39:48.159
I don't know, maybe like 10 different applications
00:39:48.159 --> 00:39:52.019
now for stuff that we use internally. So like
00:39:52.019 --> 00:39:54.000
there are certain processes that I think are
00:39:54.000 --> 00:39:56.659
better used where I could build a mini app and
00:39:56.659 --> 00:40:00.679
vibe code it using a replet. And then we use
00:40:00.679 --> 00:40:03.059
it internally for like certain automations and
00:40:03.059 --> 00:40:05.340
systems and processes for our team internally.
00:40:06.059 --> 00:40:07.980
But then also I built a few customer -facing
00:40:07.980 --> 00:40:10.260
ones. So we've got this insights tool where you
00:40:10.260 --> 00:40:13.820
plug in your company domain, and it'll generate
00:40:13.820 --> 00:40:18.239
this full report of your differentiators, your
00:40:18.239 --> 00:40:21.219
positioning. We'll pull a list of companies that
00:40:21.219 --> 00:40:23.420
you should target. We'll pull a list of people
00:40:23.420 --> 00:40:25.420
and their email addresses that you could go after.
00:40:25.539 --> 00:40:27.380
We'll write some prospecting messaging that you
00:40:27.380 --> 00:40:29.960
could take and start to tweak. We'll pull together
00:40:29.960 --> 00:40:32.019
discovery questions that you could start to ask
00:40:32.019 --> 00:40:34.059
in your sales calls. And it does all of this
00:40:34.059 --> 00:40:36.570
from simply your domain. Right. That's kind of
00:40:36.570 --> 00:40:38.309
like the lead magnet we built. Built somewhere
00:40:38.309 --> 00:40:41.969
around finding emails for people, built a quota
00:40:41.969 --> 00:40:43.889
calculator. So if you want to understand what
00:40:43.889 --> 00:40:46.070
you have to do to hit your quota, you just plug
00:40:46.070 --> 00:40:48.969
in a couple metrics into the calculator. It builds
00:40:48.969 --> 00:40:50.989
it for you. But it's like a really cool front
00:40:50.989 --> 00:40:55.190
-end UI app that I built in 45 minutes in Replit.
00:40:55.389 --> 00:40:58.349
So when you think about this from a marketing
00:40:58.349 --> 00:41:01.539
perspective... like what are the things that
00:41:01.539 --> 00:41:04.179
would make your prospect or your customers lives
00:41:04.179 --> 00:41:06.219
easier what are tools things that they could
00:41:06.219 --> 00:41:09.099
leverage build them and then give it to them
00:41:09.099 --> 00:41:11.179
and like that is going to be a much better lead
00:41:11.179 --> 00:41:14.460
magnet than the latest white paper or you know
00:41:14.460 --> 00:41:17.400
poster or whatever that every other company is
00:41:17.400 --> 00:41:20.480
doing i love that and i think the other thing
00:41:20.480 --> 00:41:23.119
that you could probably do once you start really
00:41:23.119 --> 00:41:26.559
expanding this out is because if you can automate
00:41:26.559 --> 00:41:30.159
this process in clay And you can start to do
00:41:30.159 --> 00:41:32.960
this at scale and see what tools resonate and
00:41:32.960 --> 00:41:35.480
what don't. Like you could send 10 very similar
00:41:35.480 --> 00:41:38.619
companies 10 different tools. Because you can
00:41:38.619 --> 00:41:40.539
just automate it. It still takes a day and you
00:41:40.539 --> 00:41:42.920
don't do most of the work. And see which tools
00:41:42.920 --> 00:41:46.219
resonate. I read a really interesting use case
00:41:46.219 --> 00:41:49.059
that I don't
think we can do in our industry.
00:41:49.699 --> 00:41:52.860
Maybe we could actually. But the concept was
00:41:52.860 --> 00:41:58.239
spin up business ideas through JetGPT. or some
00:41:58.239 --> 00:42:02.119
other model then create ads for those business
00:42:02.119 --> 00:42:06.360
ideas and promote them on facebook meta for and
00:42:06.360 --> 00:42:09.639
just spend like ten dollars per idea and see
00:42:09.639 --> 00:42:11.699
which ones get the most engagement and that's
00:42:11.699 --> 00:42:15.099
what then guides you in terms of what products
00:42:15.099 --> 00:42:16.900
you should develop but like your market research
00:42:16.900 --> 00:42:19.159
becomes not asking what people want but actually
00:42:19.159 --> 00:42:21.599
just advertising to them like this thing exists
00:42:21.599 --> 00:42:25.219
and if they try and buy it yeah and then if they
00:42:25.219 --> 00:42:27.440
try buy it then you build it And of course, I
00:42:27.440 --> 00:42:30.340
think if you're doing that with software, well,
00:42:30.519 --> 00:42:34.139
the build capabilities are now very flexible,
00:42:34.179 --> 00:42:36.760
very fast. So you could probably spin up the
00:42:36.760 --> 00:42:40.599
software product in quite, you know, not much
00:42:40.599 --> 00:42:42.980
time, really. And non -technical people can do
00:42:42.980 --> 00:42:45.139
it, right? I think if you're a marketer, you
00:42:45.139 --> 00:42:47.039
should learn how to use these tools because then
00:42:47.039 --> 00:42:49.199
you don't need an engineer to do this stuff for
00:42:49.199 --> 00:42:51.860
you anymore. Yeah, a friend of mine... You can
00:42:51.860 --> 00:42:53.800
do it. In fact, you know this because Nick went
00:42:53.800 --> 00:42:56.619
on this. We went on Hack Week. Oh, yeah, yeah,
00:42:56.739 --> 00:42:58.940
yeah. I saw your post about it, and I'm like,
00:42:59.000 --> 00:43:01.559
Nick, go do this. Yeah, yeah. It was really cool,
00:43:01.699 --> 00:43:04.360
and I got a huge amount of value out of it. But
00:43:04.360 --> 00:43:07.380
a couple of months later, the single most valuable
00:43:07.380 --> 00:43:11.760
thing was carving some time out to get creative
00:43:11.760 --> 00:43:14.440
and play with. I was using Replit as well, but
00:43:14.440 --> 00:43:16.619
now I'm more using Lovable, but they're basically
00:43:16.619 --> 00:43:18.340
the same thing, aren't they, pretty much? Anyway,
00:43:18.440 --> 00:43:19.960
I say that, but I think it's probably still true
00:43:19.960 --> 00:43:24.300
at this point. To just play with stuff and try
00:43:24.300 --> 00:43:27.260
and build stuff for me. It's almost like for
00:43:27.260 --> 00:43:29.880
the right organizations with the right people,
00:43:29.960 --> 00:43:31.739
maybe you don't have to do this across your whole
00:43:31.739 --> 00:43:33.619
team. Maybe you have to figure out those people
00:43:33.619 --> 00:43:35.900
you mentioned earlier, Harrison, who are a bit
00:43:35.900 --> 00:43:37.679
nerdier, quite like the idea of playing with
00:43:37.679 --> 00:43:39.719
this stuff. It's trying to apply some sort of
00:43:39.719 --> 00:43:43.820
Google 20 % time. Like one day a week, you look
00:43:43.820 --> 00:43:46.639
at some of the work that you're doing and you
00:43:46.639 --> 00:43:49.199
try and think about what front end applications
00:43:49.199 --> 00:43:53.619
you could build using Lovable or whatever to
00:43:53.619 --> 00:43:56.420
help. um you do your you and your team do your
00:43:56.420 --> 00:43:58.659
work better or even more speculatively what cool
00:43:58.659 --> 00:44:01.320
stuff could we do now you've got this you'd use
00:44:01.320 --> 00:44:02.780
that one day a week to go and figure out what
00:44:02.780 --> 00:44:05.800
that cool stuff could be yes i think every company
00:44:05.800 --> 00:44:08.280
should do this i'll give you two examples of
00:44:08.280 --> 00:44:11.659
companies that have done this there is uh if
00:44:11.659 --> 00:44:14.519
you know monday .com monday yeah like there's
00:44:14.519 --> 00:44:18.329
like a project management tool. They had a complete
00:44:18.329 --> 00:44:20.869
whole week. So where some companies might do
00:44:20.869 --> 00:44:22.510
an offsite for like a training or something,
00:44:22.590 --> 00:44:24.590
they did a whole week, flew everybody in the
00:44:24.590 --> 00:44:27.230
same place. And they did basically an AI hackathon
00:44:27.230 --> 00:44:31.909
where everybody figured out and built like real
00:44:31.909 --> 00:44:35.329
AI tools that they could use in their day -to
00:44:35.329 --> 00:44:38.579
-day jobs. and like i think it does so many it
00:44:38.579 --> 00:44:40.340
does so many things one it just gets everybody
00:44:40.340 --> 00:44:42.760
talking about this stuff thinking about it using
00:44:42.760 --> 00:44:47.739
it and then um also like you have if you if you
00:44:47.739 --> 00:44:49.880
spend the time to do this they probably came
00:44:49.880 --> 00:44:52.440
out with like 10 to 20 different ai tools that
00:44:52.440 --> 00:44:54.199
are now going to make their like make them more
00:44:54.199 --> 00:44:56.699
efficient or more effective in their jobs and
00:44:56.699 --> 00:45:00.909
it's like i don't know how This was just never
00:45:00.909 --> 00:45:03.210
possible before. And so if you're not thinking
00:45:03.210 --> 00:45:06.250
in this way, it's just so hard for me to think
00:45:06.250 --> 00:45:08.030
about not thinking in this way, I guess, right?
00:45:08.110 --> 00:45:10.349
It's like, we do this stuff all day, every day.
00:45:11.389 --> 00:45:13.250
Luckily, we're a small team, we're able to do
00:45:13.250 --> 00:45:17.750
that kind of stuff. But yeah, just so much value
00:45:17.750 --> 00:45:20.690
that you can unlock. Yeah, I mean, there's a
00:45:20.690 --> 00:45:23.809
speculative element to it, right? It's one of
00:45:23.809 --> 00:45:26.809
the frustrating things about the current tools,
00:45:26.949
--> 00:45:28.550
although again, they're getting better all the
00:45:28.550 --> 00:45:32.590
time, is they break a lot like something that
00:45:32.590 --> 00:45:36.409
not every output i get is a 45 minute you know
00:45:36.409 --> 00:45:39.090
idea to execution everything works and one of
00:45:39.090 --> 00:45:40.570
the things i found really frustrating during
00:45:40.570 --> 00:45:45.369
the hat week was i think of a new feature so
00:45:45.369 --> 00:45:48.289
i'd ask replit to add the new feature to the
00:45:48.289 --> 00:45:49.869
platform i'd already built and it would break
00:45:49.869 --> 00:45:51.670
three of the old features because it's almost
00:45:51.670 --> 00:45:53.610
like it would forget that those features existed
00:45:53.610 --> 00:45:56.010
or duplicate a database that already existed
00:45:56.010 --> 00:45:58.500
once and now Which version of the database is
00:45:58.500 --> 00:46:00.679
being updated only fuels one feature and not
00:46:00.679 --> 00:46:03.780
the other. So, you know, I don't want to give
00:46:03.780 --> 00:46:06.440
the impression that you can go ask it to build
00:46:06.440 --> 00:46:08.239
you a tool, the tool will be perfect. It's exactly
00:46:08.239 --> 00:46:10.900
the same as using AI in any other context. You're
00:46:10.900 --> 00:46:12.559
going to have to have a conversation. Something
00:46:12.559 --> 00:46:14.320
that you think is going to take 10 minutes is
00:46:14.320 --> 00:46:16.480
probably going to take three hours and be quite
00:46:16.480 --> 00:46:18.480
frustrating along the way. But if you stick at
00:46:18.480 --> 00:46:20.619
it, you will get something really cool that you
00:46:20.619 --> 00:46:23.679
never could have done otherwise. So you need
00:46:23.679 --> 00:46:26.309
that mindset. And I think sometimes... Yeah,
00:46:26.349 --> 00:46:28.449
you need that patience and you need, I think
00:46:28.449 --> 00:46:30.070
you need to have genuine interest in trying to
00:46:30.070 --> 00:46:32.130
build something cool. You probably need to be
00:46:32.130 --> 00:46:33.849
a bit of a builder, really. Otherwise, you're
00:46:33.849 --> 00:46:36.030
probably not going to stick in it. But I can
00:46:36.030 --> 00:46:37.789
see as a business as well that you might be like,
00:46:37.889 --> 00:46:42.190
so I need to send someone off one day a week.
00:46:42.210 --> 00:46:43.630
They're going to play with this stuff, but there
00:46:43.630 --> 00:46:45.889
might not be any outputs because maybe it doesn't
00:46:45.889 --> 00:46:48.309
work. And not all businesses are going to be
00:46:48.309 --> 00:46:49.809
open to that. Not all businesses are going to
00:46:49.809 --> 00:46:51.769
have the flex in their business model to do it.
00:46:51.909 --> 00:46:54.500
But I think if you can, you should, right? yeah
00:46:54.500 --> 00:46:59.139
I definitely agree um the other company I was
00:46:59.139 --> 00:47:02.940
referencing that does this um they have a Friday
00:47:02.940 --> 00:47:06.480
afternoon where the team five codes so it's sort
00:47:06.480 --> 00:47:09.860
of like this optional vibe coding session on
00:47:09.860 --> 00:47:12.820
Friday afternoons so it's like it's a Friday
00:47:12.820 --> 00:47:15.920
afternoon half the people are probably like not
00:47:15.920 --> 00:47:18.039
working at your company anyway and depending
00:47:18.039 --> 00:47:21.059
on how big the company is so like Give them something
00:47:21.059 --> 00:47:23.460
fun to do and have some outputs from it or something.
00:47:24.039 --> 00:47:29.780
I also just think it's very short -sighted to
00:47:29.780 --> 00:47:33.000
probably not be thinking about how your team
00:47:33.000 --> 00:47:34.840
or company could leverage AI more effectively.
00:47:35.659 --> 00:47:38.380
So while there's probably companies out there
00:47:38.380 --> 00:47:41.019
that will say not to use AI in these types of
00:47:41.019 --> 00:47:45.139
ways, I personally think it's short -sighted.
00:47:46.259 --> 00:47:48.500
Yeah, I agree. I mean, I think the other thing
00:47:48.500 --> 00:47:51.500
is it's very easy to say we're going to bring
00:47:51.500 --> 00:47:55.059
AI into how we work, you know, as a senior leader,
00:47:55.159 --> 00:47:57.719
maybe a CEO in the executive team, maybe the
00:47:57.719 --> 00:47:59.360
board are like, what are you doing with AI? And
00:47:59.360 --> 00:48:01.599
you're like, we're going to do some stuff. But
00:48:01.599 --> 00:48:05.019
it's like all bets. Like, it's a bet. You have
00:48:05.019 --> 00:48:07.179
to invest in it. Some stuff's not going to work.
00:48:07.360 --> 00:48:09.989
You have to be patient. you can't throw it out
00:48:09.989 --> 00:48:12.269
like after like two weeks of trying it because
00:48:12.269 --> 00:48:13.590
you didn't quite get what you needed you have
00:48:13.590 --> 00:48:15.130
to figure out what's not working and improve
00:48:15.130 --> 00:48:18.269
it and all that stuff and i don't think you can
00:48:18.269 --> 00:48:20.710
try and tie a metric to this and i think that's
00:48:20.710 --> 00:48:23.449
the thing that people don't like a lot of execs
00:48:23.449 --> 00:48:24.989
or a lot of leadership don't understand it's
00:48:24.989 --> 00:48:28.349
like you can do things that don't always have
00:48:28.349 --> 00:48:33.609
an roi and a metric tied to them like i i This
00:48:33.609 --> 00:48:35.969
is one of the things I love most about what I'm
00:48:35.969 --> 00:48:38.469
doing now, kind of my own company, is that I
00:48:38.469 --> 00:48:41.090
don't have to, if I think an idea is cool, I'm
00:48:41.090 --> 00:48:43.769
just going to do it. I don't have to think about
00:48:43.769 --> 00:48:46.010
what's the ROI for this? What's the metric? What
00:48:46.010 --> 00:48:49.070
is the whatever? I'm going to go do it, see if
00:48:49.070 --> 00:48:51.230
it works. And if it works, awesome. And I'll
00:48:51.230 --> 00:48:53.409
use qualitative feedback just to sometimes determine
00:48:53.409 --> 00:48:57.170
if it works, right? And so I think there's oftentimes...
00:48:57.690 --> 00:49:00.170
when organizations get larger and everything
00:49:00.170 --> 00:49:02.369
has to be metric driven and everything has to
00:49:02.369 --> 00:49:04.809
have a metric and a number tied to it it makes
00:49:04.809 --> 00:49:08.630
it really hard to do fun interesting things what's
00:49:08.630 --> 00:49:11.090
that saying it's not everything that counts can
00:49:11.090 --> 00:49:14.010
be measured or something everything that measured
00:49:14.010 --> 00:49:16.269
will get done or something like that but then
00:49:16.269 --> 00:49:18.250
like but then there's like the flip side which
00:49:18.250 --> 00:49:19.630
is kind of closer i think for what you're saying
00:49:19.630 --> 00:49:22.769
which is like if you can't measure it yeah i
00:49:22.769 --> 00:49:25.329
don't know what the actual one is but yeah but
00:49:25.329 --> 00:49:27.510
obviously like the sentiment is Just because
00:49:27.510 --> 00:49:29.809
you can't set clear measurables around it doesn't
00:49:29.809 --> 00:49:31.429
mean it's not important. And I think one of those
00:49:31.429 --> 00:49:33.650
things that happens, there's a couple of things
00:49:33.650 --> 00:49:35.929
that are interesting. What you learn along the
00:49:35.929 --> 00:49:38.570
way is intangibly valuable, but still can be
00:49:38.570 --> 00:49:41.789
very valuable. Plus, I think the nature of this
00:49:41.789 --> 00:49:44.349
is quite iterative. And what you get at the end
00:49:44.349 --> 00:49:46.750
might not be what you envisaged at the beginning,
00:49:46.809 --> 00:49:49.829
but it's actually 10x better than what you envisaged.
00:49:49.869 --> 00:49:52.570
Because by going through the process, you come
00:49:52.570 --> 00:49:54.269
up against a problem and you have a new idea.
00:49:54.329 --> 00:49:56.780
So you try a new thing. So it's one of those
00:49:56.780 --> 00:50:00.300
very few places where the old adage, start with
00:50:00.300 --> 00:50:02.480
the end in mind, probably doesn't help you out
00:50:02.480 --> 00:50:04.280
that much because you're trying to be creative
00:50:04.280 --> 00:50:06.679
and solve problems in a creative way. And I think
00:50:06.679 --> 00:50:09.019
as long as you have an idea directionally of
00:50:09.019 --> 00:50:11.719
where you're trying to go, I really wish we could
00:50:11.719 --> 00:50:15.460
make our content production process better. Then,
00:50:15.480 --> 00:50:17.219
well, what does that mean? Then you can start
00:50:17.219 --> 00:50:19.079
exploring it and trying different stuff. And
00:50:19.079 --> 00:50:21.440
maybe that's the best way to think about it.
00:50:22.099 --> 00:50:24.610
Yeah, I love that. It's like if you don't play
00:50:24.610 --> 00:50:26.889
with it, you can't figure it out. This is it.
00:50:28.289 --> 00:50:30.050
Right, so a couple of other questions. Have a
00:50:30.050 --> 00:50:31.409
look at what we've got on here on our little
00:50:31.409 --> 00:50:36.949
list. You've talked a bit about clay. Are there
00:50:36.949 --> 00:50:39.449
any other tools that you rely on day in, day
00:50:39.449 --> 00:50:44.150
out when using AI? Well, I think clay is more
00:50:44.150 --> 00:50:46.429
of like the orchestrator of a lot of these things.
00:50:47.110 --> 00:50:52.480
So obviously, like Claude. ChatGPT are core elements
00:50:52.480 --> 00:50:57.960
as well. There are Canva, if we're thinking about
00:50:57.960 --> 00:51:00.480
designing things, Canva is really cool for stuff
00:51:00.480 --> 00:51:04.219
like that. From a messaging perspective, there's
00:51:04.219 --> 00:51:06.099
a really cool company out there called Octave,
00:51:06.239 --> 00:51:09.320
which is where you can create these hubs around
00:51:09.320 --> 00:51:11.480
messaging and create these different messaging
00:51:11.480 --> 00:51:15.679
agents, which is really cool. Airtable is doing
00:51:15.679 --> 00:51:17.500
some cool stuff. Actually, a lot of these project
00:51:17.500 --> 00:51:19.039
management tools are doing some really cool stuff
00:51:19.039 --> 00:51:21.809
with AI. HubSpot's put some really cool stuff
00:51:21.809 --> 00:51:27.070
out with AI. Other tools, I mean, like HeyGen
00:51:27.070 --> 00:51:33.550
obviously is cool. Cursor, that's probably too
00:51:33.550 --> 00:51:35.670
advanced for a lot of people here, but like Cursor
00:51:35.670 --> 00:51:37.670
is a really cool way where you can, it's like
00:51:37.670 --> 00:51:40.610
the advanced version of Replit or Lovable, where
00:51:40.610 --> 00:51:42.210
it's a little bit more engineering heavy, but
00:51:42.210 --> 00:51:43.789
you can do some really awesome stuff with it.
00:51:44.889 --> 00:51:47.389
Yeah, and then the Replits, the Lovables, live
00:51:47.389 --> 00:51:50.239
voting apps. And I'm sure I'm missing something.
00:51:51.159 --> 00:51:53.880
There are loads of different stuff. Oh, Super
00:51:53.880 --> 00:51:55.860
Whisper. Everybody needs to get Super Whisper.
00:51:56.340 --> 00:51:59.539
It's a tool you use to transcribe, but it's a
00:51:59.539 --> 00:52:02.199
different one. So it's like voice to text. It's
00:52:02.199 --> 00:52:04.920
incredible. I use it all the time. I don't write
00:52:04.920 --> 00:52:08.800
prompts to AI anymore. I brain dump speaking
00:52:08.800 --> 00:52:12.059
to it. And it goes so much faster. And it's way
00:52:12.059 --> 00:52:13.980
more effective than trying to type everything.
00:52:15.280 --> 00:52:19.139
And yeah, it's a really good. uh voice to text
00:52:19.139 --> 00:52:22.679
tool yeah typing is where you we've talked about
00:52:22.679 --> 00:52:24.699
this before typing's been dead for a long time
00:52:24.699 --> 00:52:28.719
to me and um i've even got a local large language
00:52:28.719 --> 00:52:32.219
model on my phone now where my keyboard has a
00:52:32.219 --> 00:52:35.179
built -in transcription model but it's on device
00:52:35.179 --> 00:52:37.539
so i you know because a lot of things you might
00:52:37.539 --> 00:52:40.360
speak type on your keyboard you don't want it
00:52:40.360 --> 00:52:43.000
to come off your phone right um but that's pretty
00:52:43.000 --> 00:52:46.219
fast considering it's running on You know, Samsung,
00:52:46.519 --> 00:52:49.579
it's 23 or something. Um, the other thing is
00:52:49.579 --> 00:52:54.139
faster. It'd be on an iPhone. Probably slower
00:52:54.139 --> 00:52:57.800
these days. AI is not Apple's strong suit so
00:52:57.800 --> 00:53:01.019
far. We'll see if they improve over time. That's
00:53:01.019 --> 00:53:03.559
very true. I'm recording this on a Mac, so I'll
00:53:03.559 --> 00:53:06.119
just navigate to where the best platform is for
00:53:06.119 --> 00:53:07.760
the thing I'm trying to do at the time. There
00:53:07.760 --> 00:53:10.519
you go. The thing I really love, I think this
00:53:10.519 --> 00:53:12.239
might be boring for podcast listeners because
00:53:12.239 --> 00:53:14.079
I might have said this last week, but I'm using
00:53:14.079 --> 00:53:15.880
a tool called VoiceSync at the moment, and you
00:53:15.880 --> 00:53:19.659
can choose what model you use, and I'm using
00:53:19.659 --> 00:53:24.469
NVIDIA's new... like super fine -tuned tiny voice
00:53:24.469 --> 00:53:28.090
model and if even if i speak a paragraph it appears
00:53:28.090 --> 00:53:32.909
instantly but cool um and that that i really
00:53:32.909 --> 00:53:35.070
didn't think that latency was going to be that
00:53:35.070 --> 00:53:38.349
important to me but it really is like if i use
00:53:38.349 --> 00:53:41.050
a better model and i have to wait a second or
00:53:41.050 --> 00:53:43.710
two to get it appear on the screen i'd rather
00:53:43.710 --> 00:53:48.429
have instant transcription and fix one or two
00:53:48.429 --> 00:53:50.719
words Because they can move that much faster.
00:53:50.820 --> 00:53:52.900
It's really unbelievable. Yeah, that's really
00:53:52.900 --> 00:53:57.840
cool. I like it. Tell me, Harrison, about an
00:53:57.840 --> 00:54:01.099
AI workflow or tool or something that you thought
00:54:01.099 --> 00:54:03.940
would be really cool and work really well. And
00:54:03.940 --> 00:54:06.619
then it just didn't. And what did you learn?
00:54:07.320 --> 00:54:14.539
I tried to create a like super app myself. So
00:54:14.539 --> 00:54:16.860
I tried to build, I mean, this is still in progress.
00:54:17.039 --> 00:54:19.699
It still could come out one day. um where essentially
00:54:19.699 --> 00:54:22.679
it would be like the only app that we would need
00:54:22.679 --> 00:54:25.579
to run our whole agency so everything that we
00:54:25.579 --> 00:54:27.139
would do and so i was trying to find the whole
00:54:27.139 --> 00:54:29.079
thing building all of these different complex
00:54:29.079 --> 00:54:33.179
workflows working with databases and um it proved
00:54:33.179 --> 00:54:35.659
to be very hard as a non -technical person to
00:54:35.659 --> 00:54:38.599
work through like the slog of bugs and different
00:54:38.599 --> 00:54:41.840
challenges that would come up and so what i have
00:54:41.840 --> 00:54:45.519
like now deferred to is like simple small apps
00:54:45.519 --> 00:54:48.400
that solve very specific problems and then can
00:54:48.400 --> 00:54:50.500
make it in an hour or two, and then it's live.
00:54:50.659 --> 00:54:53.659
People can use it. And it's a lot easier than
00:54:53.659 --> 00:54:56.880
trying to build some gigantic piece of software
00:54:56.880 --> 00:55:00.260
that I don't have the technical expertise to
00:55:00.260 --> 00:55:03.679
update and maintain on a regular basis and just
00:55:03.679 --> 00:55:06.760
get it constantly working. So when you think
00:55:06.760 --> 00:55:08.699
about, this is kind of where I come back to this
00:55:08.699 --> 00:55:11.400
idea for people trying to get started with AI.
00:55:12.000 --> 00:55:15.960
is come up with something small simple easy with
00:55:15.960 --> 00:55:19.099
like defined outcomes that you can build and
00:55:19.099 --> 00:55:22.019
use right away and that will be like a nice win
00:55:22.019 --> 00:55:24.579
and like think about those things before you
00:55:24.579 --> 00:55:28.219
try and build something complex yeah that's that's
00:55:28.219 --> 00:55:30.380
a great story i've experienced that as well i
00:55:30.380 --> 00:55:32.099
got a couple of questions for you on that the
00:55:32.099 --> 00:55:34.739
first thing i just wanted to say though was and
00:55:34.739 --> 00:55:36.099
it's going to contradict what we said earlier
00:55:36.099 --> 00:55:38.679
so hopefully it still makes sense but because
00:55:38.679 --> 00:55:41.710
when i'm working with lovable or replit If I
00:55:41.710 --> 00:55:44.110
have an idea and add another feature, it breaks
00:55:44.110 --> 00:55:46.630
an old feature. One of the things I learned is
00:55:46.630 --> 00:55:49.110
sometimes it's just better to start again with
00:55:49.110 --> 00:55:51.789
a longer prompt that clearly outlines all the
00:55:51.789 --> 00:55:54.789
features you want the final product to have so
00:55:54.789 --> 00:55:57.769
that it builds them all in one go. Because what
00:55:57.769 --> 00:56:00.070
it's been like for me is if it's like dumps of
00:56:00.070 --> 00:56:03.630
like ice cream one on top of the other, like
00:56:03.630 --> 00:56:05.690
the whole thing is really flimsy and melts really
00:56:05.690 --> 00:56:08.469
easily. Yeah, yeah. Like I need to decide what
00:56:08.469 --> 00:56:11.250
that. everything is give it to the tool in one
00:56:11.250 --> 00:56:13.869
go and get a thing out but my question for you
00:56:13.869 --> 00:56:17.750
is have you had any experience like chaining
00:56:17.750 --> 00:56:21.130
your micro tools like can one tool use another
00:56:21.130 --> 00:56:26.239
tool or does that work yes You can. And the way
00:56:26.239 --> 00:56:29.099
you do that is you
would tell the tool to create
00:56:29.099 --> 00:56:32.340
a set of public API keys. Again, kind of coming
00:56:32.340 --> 00:56:35.699
back to APIs. So you create a tool that then
00:56:35.699 --> 00:56:38.559
has public API keys that can run certain functions
00:56:38.559 --> 00:56:42.099
in that tool. And then one tool can then send
00:56:42.099 --> 00:56:46.019
API requests to the other tool to then execute
00:56:46.019 --> 00:56:48.360
a function and return it back to that initial
00:56:48.360 --> 00:56:53.190
tool. So an example is... Let's say I wanted
00:56:53.190 --> 00:56:55.610
to build a database of all the leads that we
00:56:55.610 --> 00:56:58.329
might want to target. Well, I can build this
00:56:58.329 --> 00:57:00.309
database. I can enable this database to have
00:57:00.309 --> 00:57:04.190
specific API calls that I might want to enrich
00:57:04.190 --> 00:57:06.550
a lead, to go and get specific fields from a
00:57:06.550 --> 00:57:09.769
particular lead. Then in another app, I might
00:57:09.769 --> 00:57:12.349
have something where I want to, if I plug in
00:57:12.349 --> 00:57:15.349
a person's name and their company, I want it
00:57:15.349 --> 00:57:18.780
to go check this database to see if We have the
00:57:18.780 --> 00:57:20.820
email already for them. And it could easily go
00:57:20.820 --> 00:57:23.320
and do that by simply having like these API calls
00:57:23.320 --> 00:57:26.519
set up. Or in the example of like that insights
00:57:26.519 --> 00:57:29.239
tool I mentioned is I leveraged probably like
00:57:29.239 --> 00:57:32.960
seven different APIs in that tool to then get
00:57:32.960 --> 00:57:34.960
all of the data that we want to show up in that
00:57:34.960 --> 00:57:38.019
tool itself. Yeah, there's definitely a systems
00:57:38.019 --> 00:57:40.039
architecture skill to this in terms of being
00:57:40.039 --> 00:57:42.260
able to break a process down into its constituent
00:57:42.260 --> 00:57:45.679
steps, thinking about what information needs
00:57:45.679 --> 00:57:48.500
to come from where. how it needs to be transformed
00:57:48.500 --> 00:57:51.920
or used step to step i wish it was a strength
00:57:51.920 --> 00:57:53.719
of mine because i really feel like i could build
00:57:53.719 --> 00:57:56.280
better things and more clearly articulate to
00:57:56.280 --> 00:57:59.400
the ai what i want and maybe i can take my fuzzy
00:57:59.400 --> 00:58:01.139
ideas and turn them into system architecture
00:58:01.139 --> 00:58:04.219
it does and that was going to recommend so you
00:58:04.219 --> 00:58:06.260
mentioned at the beginning when you're talking
00:58:06.260 --> 00:58:08.519
about like how do you build these like prompts
00:58:08.519 --> 00:58:10.820
to get all the features built one thing that
00:58:10.820 --> 00:58:14.360
you can do is you can use voice to text and brain
00:58:14.360 --> 00:58:18.260
dump your idea into a chat GPT, put it on thinking
00:58:18.260 --> 00:58:20.639
mode and say, build me a product requirements
00:58:20.639 --> 00:58:23.980
document for this product. And then it will go
00:58:23.980 --> 00:58:25.400
through and think through all the architecture
00:58:25.400 --> 00:58:27.219
stuff and think through all of the different
00:58:27.219 --> 00:58:29.739
features and user flows. And then you can edit
00:58:29.739 --> 00:58:34.619
it more. And then you give that prompt to replant
00:58:34.619 --> 00:58:37.500
and say, build this. Thanks. And then it will
00:58:37.500 --> 00:58:39.920
go and build that prompt and then build the product.
00:58:41.289 --> 00:58:43.530
And you still have to debug along the way, but
00:58:43.530 --> 00:58:45.710
you're creating much better. You're using it
00:58:45.710 --> 00:58:48.710
as like a meta prompt. That's my Saturday and
00:58:48.710 --> 00:58:52.650
Sunday sorted, Harrison. As we sort of drift
00:58:52.650 --> 00:58:54.429
towards the end of our conversation here, a couple
00:58:54.429 --> 00:58:57.429
of questions more about the future. So just for
00:58:57.429 --> 00:59:00.269
you, looking ahead sort of over the next 12 months,
00:59:00.590 --> 00:59:03.250
what do you think is overhyped at the moment
00:59:03.250 --> 00:59:06.789
in terms of its impact on your role? And conversely,
00:59:06.789 --> 00:59:08.289
what do you think is actually going to quietly
00:59:08.289 --> 00:59:12.829
change how... sales pros marketers do their work
00:59:12.829 --> 00:59:16.110
take the second one first um the second one i
00:59:16.110 --> 00:59:22.250
think that uh sdrs and prospecting is going to
00:59:22.250 --> 00:59:26.150
become almost obsolete and be handled by a singular
00:59:26.150 --> 00:59:29.190
human being that is incredible at using all of
00:59:29.190 --> 00:59:31.369
these different tools and systems so somebody
00:59:31.369 --> 00:59:33.849
who can basically become like an architect of
00:59:33.849 --> 00:59:37.139
your like go -to -market processes so they can
00:59:37.139 --> 00:59:39.739
set up systems to listen for different signals
00:59:39.739 --> 00:59:43.420
news hiring promotions publications that come
00:59:43.420 --> 00:59:45.219
out all of that type of stuff they listen for
00:59:45.219 --> 00:59:48.239
it comes into a tool like clay things get filtered
00:59:48.239 --> 00:59:51.360
messaging gets drafted messaging gets sent inbound
00:59:51.360 --> 00:59:54.679
leads come into your system there's literally
00:59:54.679 --> 00:59:57.800
no excuse anymore to not follow up on leads right
00:59:57.800 --> 00:59:59.179
i think one of the questions you're going to
00:59:59.179 --> 01:00:00.980
ask is about lead scoring i think league swing
01:00:00.980 --> 01:00:04.440
is a bunch of bogus and like follow up with everybody.
01:00:04.659 --> 01:00:06.380
There's no excuse to not follow up with anybody.
01:00:06.679 --> 01:00:09.119
There was always never an excuse, but salespeople
01:00:09.119 --> 01:00:10.880
had time that they had to spend on this stuff.
01:00:11.039 --> 01:00:14.099
So as marketers, automate that process. Build
01:00:14.099 --> 01:00:16.780
and write personalized custom one -to -one messages
01:00:16.780 --> 01:00:18.940
to every single lead that get followed up with
01:00:18.940 --> 01:00:21.440
within five minutes. Guaranteed, you will convert
01:00:21.440 --> 01:00:25.500
infinitely more people into meetings, into opportunities.
01:00:26.239 --> 01:00:29.380
I think these things are just going to be all
01:00:29.380 --> 01:00:31.199
centralized around a person who's really good
01:00:31.199 --> 01:00:35.070
at this stuff. And what that's going to do is
01:00:35.070 --> 01:00:37.809
free up the sales reps to focus on their top
01:00:37.809 --> 01:00:40.349
tier accounts, right? Where you still want that
01:00:40.349 --> 01:00:42.349
human element, right? So there's like this other
01:00:42.349 --> 01:00:44.690
component, which you think of, how do I prioritize
01:00:44.690 --> 01:00:47.329
all of the accounts in my territory? How do I
01:00:47.329 --> 01:00:49.909
prioritize everything within our total addressable
01:00:49.909 --> 01:00:52.409
market? With tools like Clay and these other
01:00:52.409 --> 01:00:54.789
data sources out there, you can now literally
01:00:54.789 --> 01:00:57.190
pull a list of every single company, every single
01:00:57.190 --> 01:00:59.429
person in your total addressable market. You
01:00:59.429 --> 01:01:01.860
can score them, you can rank them. pass some
01:01:01.860 --> 01:01:05.199
off to sales you own these we take this whole
01:01:05.199 --> 01:01:08.059
long tail sales is now focused and they're held
01:01:08.059 --> 01:01:10.380
accountable to everything going on at those top
01:01:10.380 --> 01:01:12.440
accounts so those top you know sorts of people
01:01:12.440 --> 01:01:15.219
that they're going after and then this like long
01:01:15.219 --> 01:01:17.659
tail gets bubbled up to sales when they're ready
01:01:17.659 --> 01:01:20.679
and so i think like architecting these systems
01:01:20.679 --> 01:01:24.179
in a way is like very much the now and future
01:01:24.179 --> 01:01:29.070
that companies should be moving towards And I
01:01:29.070 --> 01:01:30.389
just gave a couple of examples of workflows,
01:01:30.590 --> 01:01:31.989
but like there's probably infinitely more that
01:01:31.989 --> 01:01:34.190
I'm just, you know, we probably don't have time
01:01:34.190 --> 01:01:36.550
to talk through. Yeah. I love the use of the
01:01:36.550 --> 01:01:38.949
word conductor. Like sometimes people ask me,
01:01:39.010 --> 01:01:41.630
especially people earlier in their career, like
01:01:41.630 --> 01:01:44.110
how should I respond to all these things that
01:01:44.110 --> 01:01:46.010
are going on in AI? And there's a lot of advice
01:01:46.010 --> 01:01:48.289
out there and a lot of it's really good in terms
01:01:48.289 --> 01:01:51.510
of stay up to date on what, you know, different
01:01:51.510 --> 01:01:54.670
tools can do and where
AI is headed. Become an
01:01:54.670 --> 01:01:57.369
AI native and be prepared to play with all these
01:01:57.369 --> 01:01:58.989
tools exactly like we've talked about today.
01:01:59.150 --> 01:02:01.670
But one of the roles I do think is going to become
01:02:01.670 --> 01:02:05.630
critical in organizations is the conductor. The
01:02:05.630 --> 01:02:09.630
one who is able to leverage these tools to deliver
01:02:09.630 --> 01:02:13.090
different tasks, bring the outputs together to
01:02:13.090 --> 01:02:16.630
generate meaningful work and outcomes and troubleshoot
01:02:16.630 --> 01:02:18.670
and debug and improve them when they break, which
01:02:18.670 --> 01:02:22.260
I suspect they will break a lot. which is probably
01:02:22.260 --> 01:02:25.679
going to be quite a weird hybrid of subject matter
01:02:25.679 --> 01:02:29.000
expertise in the thing, like knowing what good
01:02:29.000 --> 01:02:32.119
prospecting looks like and knowing how you should
01:02:32.119 --> 01:02:35.300
combine all of these things, but also a technical
01:02:35.300 --> 01:02:38.340
capability that is associated with optimizing
01:02:38.340 --> 01:02:40.880
and improving the tools, stress testing the tools,
01:02:41.000 --> 01:02:43.519
fixing them when they break. And so that's probably,
01:02:43.579 --> 01:02:45.199
I think, where people are going to end up spending
01:02:45.199 --> 01:02:47.860
a lot of their time to train into the roles of
01:02:47.860 --> 01:02:53.079
tomorrow. I agree. I agree. Where do you think
01:02:53.079 --> 01:02:55.340
it's going to be overhyped? Why is everybody
01:02:55.340 --> 01:02:57.320
talking about this amazing right now that's going
01:02:57.320 --> 01:03:01.019
to take longer than everybody thinks? I don't
01:03:01.019 --> 01:03:04.639
really personally believe that we're going to
01:03:04.639 --> 01:03:08.659
get to this AGI type of thing anytime soon, where
01:03:08.659 --> 01:03:11.679
it's this super intelligence that's going to
01:03:11.679 --> 01:03:14.639
be able to be good at very specific things. I
01:03:14.639 --> 01:03:17.940
think we're going to have... the models will
01:03:17.940 --> 01:03:19.920
be great but they're always going to need to
01:03:19.920 --> 01:03:22.300
be fine -tuned for a period of time and so like
01:03:22.300 --> 01:03:24.099
there's all this like hype and talk about like
01:03:24.099 --> 01:03:26.300
we're going to hit AGI super intelligence is
01:03:26.300 --> 01:03:29.780
on the way and it's like baby but like in reality
01:03:29.780 --> 01:03:31.880
it's probably going to show up at point solutions
01:03:31.880 --> 01:03:34.239
and very specific workflows where that expertise
01:03:34.239 --> 01:03:37.679
comes in where you have that like level of um
01:03:37.679 --> 01:03:40.380
like the value actually is going to come from
01:03:40.380 --> 01:03:45.340
the specificity not the like PhD level GPT -5.
01:03:45.960 --> 01:03:48.579
Like, it's going to be in the, like, 27 -page
01:03:48.579 --> 01:03:51.440
document that you have where you outline exactly
01:03:51.440 --> 01:03:55.639
what to do in each scenario. And I think that
01:03:55.639 --> 01:03:58.840
that's still going to be the case. So we'll still
01:03:58.840 --> 01:04:01.280
need a lot of humans to manage and handle that
01:04:01.280 --> 01:04:05.860
stuff. Yeah, it still feels incredibly fragmented.
01:04:06.119 --> 01:04:07.599
Like, even the fact that we're talking about
01:04:07.599 --> 01:04:10.019
creating micro tools that we then have to find
01:04:10.019 --> 01:04:11.739
a clever way to make them play well together,
01:04:11.880 --> 01:04:13.239
and then most of the time they're still going
01:04:13.239 --> 01:04:17.170
to break. Yeah. That's annoying for you and I,
01:04:17.269 --> 01:04:19.989
when we spend a lot of time and are willing to
01:04:19.989 --> 01:04:23.150
invest a lot of time in playing with them. For
01:04:23.150 --> 01:04:26.070
your average marketer or salesperson and a team
01:04:26.070 --> 01:04:28.849
that may not have that time or interest and they
01:04:28.849 --> 01:04:31.130
just need it to work, it feels like we're still
01:04:31.130 --> 01:04:33.429
quite a long way away from that. So it's quite
01:04:33.429 --> 01:04:35.389
a weird time from that perspective. But then
01:04:35.389 --> 01:04:38.929
the speed that things are moving at is kind of
01:04:38.929 --> 01:04:40.690
hard to quantify. And I was listening to a podcast
01:04:40.690 --> 01:04:44.139
yesterday, Sam Morton was on the... Is it A16Z?
01:04:44.260 --> 01:04:45.980
I think that's what the investment firms call
01:04:45.980 --> 01:04:49.360
Marc Andreessen. And it was the first time I'd
01:04:49.360 --> 01:04:52.219
ever heard or the first time in recent times
01:04:52.219 --> 01:04:55.760
that I've heard someone, a senior person at an
01:04:55.760 --> 01:05:00.079
AI company or lab say, what we are seeing internally
01:05:00.079 --> 01:05:02.780
is one to two years ahead of what's in the market.
01:05:02.960 --> 01:05:06.219
Because basically it was justifying why they're
01:05:06.219 --> 01:05:08.840
making all these bets and spending all this money.
01:05:10.320 --> 01:05:12.380
in a sort of backwards way he said you know you
01:05:12.380 --> 01:05:15.940
wouldn't invest that money if you didn't see
01:05:15.940 --> 01:05:19.639
what was coming sure yeah um whereas i think
01:05:19.639 --> 01:05:21.559
that sounds like a hype message to me though
01:05:21.559 --> 01:05:24.460
i mean he's probably true i'm sure it's probably
01:05:24.460 --> 01:05:26.019
true but like he's probably out there marketing
01:05:26.019 --> 01:05:28.559
at the same time that he says that always and
01:05:28.559 --> 01:05:30.880
this is so hard now to know what you can trust
01:05:30.880 --> 01:05:34.800
probably nothing um Especially now that his avatar
01:05:34.800 --> 01:05:36.920
is on Sora 2 and there's all these different
01:05:36.920 --> 01:05:40.219
videos being made of
him. That's so funny. That's
01:05:40.219 --> 01:05:42.699
quite cool. That's quite brave of him to say,
01:05:42.820 --> 01:05:44.820
I'm going to stick my neck out. I'm going to
01:05:44.820 --> 01:05:46.780
be the first one that makes their avatar available.
01:05:47.460 --> 01:05:53.579
Yeah. When it comes to AI, what's the thing that
01:05:53.579 --> 01:05:57.039
you're most excited about? Yeah, I don't actually
01:05:57.039 --> 01:06:00.739
know. I'm already excited. I am like, this is...
01:06:00.920 --> 01:06:02.860
Like this is all like fun. Like everything is
01:06:02.860 --> 01:06:05.500
enjoyable. Like it's fun. And I think that's
01:06:05.500 --> 01:06:07.059
the thing that excites me is like, how is this
01:06:07.059 --> 01:06:08.860
going to become more fun, more entertaining,
01:06:09.000 --> 01:06:11.079
where we can tinker more, we can build more things.
01:06:11.159 --> 01:06:13.780
Like there's less limitations on what is possible.
01:06:14.079 --> 01:06:16.679
I think that's like a continuation of where we're
01:06:16.679 --> 01:06:19.260
at today. That's kind of where I'm at. It's fun.
01:06:19.380 --> 01:06:21.679
It's a fun time to be being able to play with
01:06:21.679 --> 01:06:24.900
all of this stuff. I appreciate that. So last
01:06:24.900 --> 01:06:27.889
question. What have we not talked about today
01:06:27.889 --> 01:06:29.909
that you think would have been cool to talk about?
01:06:29.989 --> 01:06:32.969
Is there anything still on your mind? We talked
01:06:32.969 --> 01:06:39.929
about a lot of stuff. Not really. I mean, I think
01:06:39.929 --> 01:06:41.289
we talked about a lot of things. I mean, maybe
01:06:41.289 --> 01:06:43.130
the only thing I would say is like, if you want
01:06:43.130 --> 01:06:47.210
any advice, help to like build out any of these
01:06:47.210 --> 01:06:49.429
different systems or tools, or you just need
01:06:49.429 --> 01:06:51.289
help building top of funnel pipeline, obviously
01:06:51.289 --> 01:06:53.849
reach out to succession .bio. Happy to help.
01:06:53.989 --> 01:06:56.630
So there's my plug. Yeah, no, I love that. I
01:06:56.630 --> 01:06:58.889
was going to ask you to give me contact details
01:06:58.889 --> 01:07:00.289
so people get in touch with you. But I think
01:07:00.289 --> 01:07:01.730
you raised a really important point. It's come
01:07:01.730 --> 01:07:03.710
up a number of times in the podcast today, which
01:07:03.710 --> 01:07:07.050
is some people love tinkering and playing and
01:07:07.050 --> 01:07:09.550
building. And some people want something that
01:07:09.550 --> 01:07:11.349
works because they've got results to deliver.
01:07:11.809 --> 01:07:14.909
And I think, and I don't think there's any, if
01:07:14.909 --> 01:07:18.769
there has been in this podcast, I haven't seen
01:07:18.769 --> 01:07:20.829
one. There's no value judgment in that. It's
01:07:20.829 --> 01:07:22.969
different people are good at different things
01:07:22.969 --> 01:07:27.030
and they enjoy. different things so for those
01:07:27.030 --> 01:07:29.530
of you that find this stuff nerdy cool and want
01:07:29.530 --> 01:07:31.570
to go play with it hopefully you feel encouragement
01:07:31.570 --> 01:07:35.289
to go play with it if you are a get stuff done
01:07:35.289 --> 01:07:38.630
outcomes i i want the outcomes of that stuff
01:07:38.630 --> 01:07:41.550
i don't want to build it myself go find people
01:07:41.550 --> 01:07:44.909
who like to build the stuff and then you can
01:07:44.909 --> 01:07:48.309
surf the waves they create to like massively
01:07:48.309 --> 01:07:50.889
amplify the impact that you can have and that's
01:07:50.889 --> 01:07:53.250
where you can work with people like Harrison
01:07:53.250 --> 01:07:56.969
on the sales side. We have a consulting and AI
01:07:56.969 --> 01:07:59.849
implementation arm at Supreme Group when it comes
01:07:59.849 --> 01:08:02.489
to more of the marketing side. There are plenty
01:08:02.489 --> 01:08:04.329
of people out there now who've been developing
01:08:04.329 --> 01:08:06.449
these skills that you can leverage to help you
01:08:06.449 --> 01:08:10.309
get better results. So on that note, thank you
01:08:10.309 --> 01:08:11.809
so much for coming on, Harrison. It's been a
01:08:11.809 --> 01:08:13.710
pleasure to hang out with you. Thank you. The
01:08:13.710 --> 01:08:16.270
podcast is back. I'm excited to be the second
01:08:16.270 --> 01:08:20.319
guest on the second iteration of the show. Pleasure's
01:08:20.319 --> 01:08:22.439
been all ours, and I'm sure I'll see you soon.
01:08:22.500 --> 01:08:24.180
Cheers, buddy. Of course. Yeah, man. Have a good
01:08:24.180 --> 01:08:27.779
one. Talk to you later. Thank you for listening
01:08:27.779 --> 01:08:30.899
to Artificially Intelligent Marketing. To stay
01:08:30.899 --> 01:08:34.000
on top of the latest trends, tips, and tools
01:08:34.000 --> 01:08:37.500
in the world of marketing AI, be sure to subscribe.
01:08:38.000 --> 01:08:40.579
We look forward to seeing you again next week.