WEBVTT
00:00:01.679 --> 00:00:09.759
Welcome to another episode of Bionic Advisor, where we take a deep dive into all the amazing pieces of tech being used by financial planning firms up and down the country.
00:00:10.000 --> 00:00:14.880
If you're a financial planner struggling to keep up on the tech treadmill, this is your home.
00:00:15.119 --> 00:00:20.640
For this episode, I'm joined by the amazing Phil Teal, who runs his own business, Teal Pen and Pixel.
00:00:20.719 --> 00:00:21.920
Thanks for joining us, Phil.
00:00:22.399 --> 00:00:23.760
Pleasure to be here.
00:00:24.879 --> 00:00:38.000
So I've known Phil for some time now, and the two of us initially bonded over a mutual love of a tool that I would say is one of the best kept secrets from financial planners.
00:00:38.320 --> 00:00:40.640
And that tool is called NA10.
00:00:41.039 --> 00:00:47.280
I've only met a handful of financial planners who have even heard of NAN, let alone use it day to day.
00:00:47.600 --> 00:00:52.640
But uh for me personally, I'm at the point now where I could not run my business without it.
00:00:52.880 --> 00:00:59.920
So I asked Phil to join me for an episode to have a chat about NA10 because he is a daily NAN user.
00:01:00.079 --> 00:01:05.519
He is building workflows within NAN on a daily basis for financial planning firms.
00:01:05.840 --> 00:01:13.359
And it's such an amazing piece of kit that I honestly think this might be the most useful episode of Bionic Advisor that we ever do.
00:01:13.680 --> 00:01:16.000
And it's just the start, so it's all downhill from here.
00:01:16.079 --> 00:01:17.439
There's nothing gonna be better than this.
00:01:17.840 --> 00:01:20.799
Um let's get into it, shall we?
00:01:21.359 --> 00:01:22.480
Let's do it.
00:01:22.799 --> 00:01:23.120
Okay.
00:01:23.519 --> 00:01:27.359
Okay, so I think first off, nice and easy.
00:01:27.599 --> 00:01:35.760
In plain English, do you just want to explain what NAN is and how you would describe it to a planner who's never touched anything like it?
00:01:36.719 --> 00:01:37.359
Yes.
00:01:37.599 --> 00:01:50.480
Um so NA10 is it's a bit like a canvas where you can drag different apps onto this canvas or workbench and connect them together.
00:01:50.640 --> 00:02:00.719
Um so the idea is you're trying to get different systems talking to each other that might not normally do that, and it's incredibly flexible and versatile.
00:02:00.959 --> 00:02:12.080
So it's going beyond um, you know, for instance, connecting something like Airtable to Google Docs, although you know that's a great type of integration.
00:02:12.560 --> 00:02:28.240
But you can connect um to various platforms and CRMs um that might seem impenetrable through, for instance, something called an HTTP request, which is a special node in NA10.
00:02:28.319 --> 00:02:30.240
Um, and it's just very powerful.
00:02:30.400 --> 00:02:32.159
And um I love it.
00:02:32.240 --> 00:02:39.840
I use it every day, and it powers a lot of my own business operations, and it's incredibly useful for advisors as well.
00:02:40.000 --> 00:02:45.039
And uh I wish more knew about it, so I'm very glad to be here to to talk about it.
00:02:45.759 --> 00:02:46.560
Definitely.
00:02:46.719 --> 00:02:57.520
Um so on um on a previous episode with Dan Graham, I uh talked about Zapia, and there's a lot of similarities, it's fair to say, between Zapia and NAN.
00:02:57.840 --> 00:03:01.919
But uh another one that I hear people talk about quite often is make.
00:03:02.560 --> 00:03:07.919
But how would you say that NAN differs from something like Zapia or Make?
00:03:09.199 --> 00:03:16.719
Yeah, so I would say actually that something like Zapia is great to get started with something like NA10.
00:03:16.879 --> 00:03:21.199
You can go straight to NA10, but Zapia is very beginner-friendly.
00:03:21.520 --> 00:03:27.280
But where NA10 really shines is on quite a few different fronts.
00:03:27.439 --> 00:03:28.960
One is uh cost.
00:03:29.360 --> 00:03:40.080
So once you start scaling up with NA10, um with Zapia, sorry, you find that it can get quite expensive because you're paying per execution.
00:03:40.240 --> 00:03:43.199
Uh, with NA10, you can self-host it.
00:03:43.360 --> 00:03:59.120
So you can put it on uh a computer that you're renting somewhere, something called a VPS, and that could be as little as eight pounds a month, and you can run it as much as you like, really within the limits of that computer.
00:03:59.680 --> 00:04:06.719
Um so as the business grows and as you add people, add workflows to it, it stays very lean.
00:04:07.120 --> 00:04:12.000
Um and uh it ties into another thing I really like about it, which is ownership.
00:04:12.159 --> 00:04:20.800
Um, you can use NA10 on the cloud, and that's where I would recommend most people start as they go to the website and just give it a free trial.
00:04:21.120 --> 00:04:31.279
Um, but when you do put it on your own rented computer, a VPS, then you are effectively it's yours because it's open source.
00:04:31.439 --> 00:04:40.160
And in fact, you can run it on your local machine if you absolutely wanted to, it's a bit more complicated to set up, but um you could do that.
00:04:40.319 --> 00:04:50.959
So um and I I would just say as well that the functionality it offers is just it's it's m far more flexible than uh Zapia.
00:04:51.279 --> 00:04:56.639
Uh some of the workflows you can build are far more complicated if you need them to be.
00:04:56.800 --> 00:05:00.879
Um so it's just a fantastic tool on so many different fronts.
00:05:01.040 --> 00:05:03.040
Uh so yeah, those are three reasons I'd say.
00:05:03.279 --> 00:05:04.959
I'm glad you touched on the the pricing.
00:05:05.040 --> 00:05:17.279
That's a it's a great point because one of the key things that they differ on is uh you know, when you build uh certainly when you build into like multi-flow stuff in Zapia, they actually calculate each step of that as a zap.
00:05:17.519 --> 00:05:24.079
So if you if you have one thing leading to another, that's one, but then if that second thing leads to a third thing, that's actually another zap.
00:05:24.319 --> 00:05:27.519
And you're you're limited, your pricing is based on how many zaps.
00:05:27.600 --> 00:05:31.600
Whereas NA10, one execution is the whole workflow, right?
00:05:31.680 --> 00:05:52.639
So if you just you can build some absolute behemoth of a workflow, even if you're on the cloud, which I think from memory, the starter on the cloud is uh 20 pounds a month, and I you get like two and a half thousand workflow executions, that's a lot of stuff to be going through for just 20 pounds a month, like 2,500 full workflows from start to finish.
00:05:52.800 --> 00:05:54.399
So that's a massive part of it.
00:05:54.639 --> 00:06:04.079
I think you're right though, for me, it's it's a little bit more flexible, it's a little bit more in the way of just it can cope with anything.
00:06:04.240 --> 00:06:11.680
I think with Zapia, you're essentially limited by what connections they've built themselves, you know.
00:06:11.759 --> 00:06:23.439
So it's if if Zapia have decided that the link between Airtable and and Gmail needs to have these triggers and these results, um that's basically all you're gonna get.
00:06:23.680 --> 00:06:36.480
Whereas really, when you get into NA10, because of how powerful the uh the actual coding is, because you you can you can get the code uh code nodes, can't you, in NA10?
00:06:36.959 --> 00:06:44.160
And that can then interpret stuff in a way that uh you would otherwise be reliant on something like Zapia to build.
00:06:44.240 --> 00:06:54.000
Whereas in NA10, you're you're much closer to the full code inside, fair to say, than like something like Zapia, which is just a little bit more plug-and-play.
00:06:54.959 --> 00:06:55.680
Absolutely.
00:06:55.920 --> 00:07:02.319
Yeah, um it you could it's very uh there's a big spectrum of users of NA10.
00:07:02.560 --> 00:07:09.839
So when I started, I was very um unaware of code and JavaScript and all these things.
00:07:10.000 --> 00:07:18.480
And it's actually been a fantastic platform for me to build my confidence in development languages and the sorts of applications I'm building now.
00:07:18.639 --> 00:07:25.839
I I would attribute a lot of it to my time on NA10 because you're learning by uh solving problems.
00:07:26.160 --> 00:07:26.720
By doing it.
00:07:26.800 --> 00:07:27.040
Yeah.
00:07:27.360 --> 00:07:38.959
Trying to figure out how do I get this code node to take this jarble, garbled up mess of something that's come from another app and and streamline it into something that another app can understand.
00:07:39.040 --> 00:07:44.720
And often you need something like a code node to do that, which which is where some of the other platforms maybe are more limited.
00:07:44.959 --> 00:07:47.519
So it is fantastic.
00:07:47.680 --> 00:07:53.120
And the cost as well, you know, obviously clawed code is all the rage right now, and I love clawed code.
00:07:53.199 --> 00:07:58.560
I use it too, but um, you know, if you're pushing that quite hard, it it does get expensive too.
00:07:58.720 --> 00:08:02.879
So I think some of the tiers are something like£100 or£100 a month.
00:08:03.040 --> 00:08:17.360
So the the great thing about NA10 is as you say, it it it's very cost effective, and you can actually talk to AI models from NA10, and you pay as you go as you do it, um, which is actually incredibly cheap.
00:08:17.519 --> 00:08:29.040
I mean, some models are costlier than others, but um you know, you're talking pennies, um, fractions of pennies for some of the calls, the API calls to these different models.
00:08:29.279 --> 00:08:31.519
And the other thing is you can call different ones.
00:08:31.680 --> 00:08:41.039
So you're not just limited to Claude, you can call open AI, perplexity, um you know, um uh all sorts, mistral.
00:08:41.600 --> 00:08:45.360
Just it's for it, you've got some sort of backups built into it as well.
00:08:45.519 --> 00:08:51.120
If one of the providers isn't working when the workflow runs, you can you can fall back to another model.
00:08:51.440 --> 00:08:53.759
So I really like it for that as well.
00:08:54.639 --> 00:08:56.639
Also the other way around as well, though, isn't it?
00:08:56.720 --> 00:09:04.000
Because I think that's a thing that puts people off when they're hearing about we're saying this this is a bit of kit that's very close to full coding.
00:09:04.240 --> 00:09:06.879
And people are like, whoa, whoa, whoa, I'm not a coder.
00:09:06.960 --> 00:09:11.440
Like the thing is you can actually get Clawed to build stuff in NAN for you.
00:09:11.600 --> 00:09:22.720
But I think, and you mentioned Clawed Code, because I do use Claude code, but the thing I still like about NAN above Claude Code is it's so visual, so it's very easy to see the flow of what's actually going on.
00:09:22.799 --> 00:09:24.960
Whereas clawed code, that's pretty much hidden.
00:09:25.039 --> 00:09:28.480
There's no real friendly interface for you to be able to see.
00:09:28.720 --> 00:09:35.279
This is going to trigger this to happen, it's gonna go down this condition path, and then it's gonna result in these potential outcomes.
00:09:35.440 --> 00:09:37.519
You can't really get that in in clawed code.
00:09:37.679 --> 00:09:42.639
So you get the best of both worlds because you can have N8N and you can still have Claude build your workflow in NAN for you.
00:09:43.519 --> 00:09:44.639
That's absolutely right.
00:09:44.720 --> 00:09:45.360
I I agree.
00:09:45.440 --> 00:09:48.559
I'm I'm quite a visual person as well, so I love it for that.
00:09:48.799 --> 00:09:55.679
And yeah, it can be quite intimidating going to Claude Code or something like that's got a command line type interface.
00:09:55.840 --> 00:09:58.080
I think a lot of people get intimidated by that.
00:09:58.240 --> 00:10:07.360
And NA10 is a lot more, you know, you just arrive at the canvas, it's a blank canvas, you have a manual trigger to start, and then just connect something to it.
00:10:07.600 --> 00:10:16.799
And you've got a workflow um as simple as that, and just build on it and see see where your imagination goes with it.
00:10:16.960 --> 00:10:18.879
Um so absolutely.
00:10:19.120 --> 00:10:24.080
Well, so um within your business, this is obviously this is what you're doing day in, day out for financial planning firms.
00:10:24.240 --> 00:10:28.320
You are building these types of workflows and and is a huge part of what you do.
00:10:28.879 --> 00:10:36.720
What's like give me some examples of some common workflows that you would build for an advice firm, some like the bread and butter use cases.
00:10:38.639 --> 00:10:44.320
So um one that's a lot of people are asking about at the moment is a meeting note-taker.
00:10:44.559 --> 00:10:52.960
So there are advisor techs that you can sign up for, which will give you a note-taker, but they can be quite costly.
00:10:53.120 --> 00:11:01.519
So I recently learned as well one of them quite significantly put up its prices, and these charge per advisor per month.
00:11:01.759 --> 00:11:06.480
So if you've got three or four advisors in your business, that's going to add up very quickly.
00:11:06.799 --> 00:11:12.799
So one workflow that I've been working on is an NA10 meeting note taker.
00:11:13.120 --> 00:11:21.919
And the wonderful thing about NA10 is that you can call the services directly that a lot of these advisor techs are actually using under the hood.
00:11:22.080 --> 00:11:23.840
Um I'm not bashing them all.
00:11:23.919 --> 00:11:34.720
I think some of them are quite interesting and useful, but um a lot of them are sort of packaging, something that you could, if you want to, build yourself.
00:11:34.960 --> 00:11:46.720
So you could have it transcribe a meeting for you after listening to the audio, and then summarize it into the structure that you want, which could then go into your CRM.
00:11:46.879 --> 00:11:48.639
So that's one example.
00:11:48.879 --> 00:11:53.360
Uh another example would be a suitability report generator.
00:11:53.919 --> 00:12:23.039
Uh so you feed it uh the fact find, you know, the various meeting notes you've got on the client, put that into a folder, and you have an A10, watch the folder, and when when all the conditions are set, it triggers and creates a brand good suitability report for you, which gets you maybe 70 or 80% of the way there, and then you go over it and check it's right, and then re-ingest it to get it almost to the finish line, and then you polish it off.
00:12:23.279 --> 00:12:24.960
So that's another useful one.
00:12:25.200 --> 00:12:28.879
Uh so and there's just so many uh that's sort of back office.
00:12:28.960 --> 00:12:49.120
I suppose there's also front office too, where um you can use it to join up your website contact form to your mailer light subscriber list, and then you've got a pipeline there to to nudge people uh along um to to book a meeting if they're not ready right there and then.
00:12:49.279 --> 00:12:51.120
So that that's something I use as well.
00:12:51.360 --> 00:13:05.759
If people download a resource from my site, um they might not want to talk straight away, and I understand that, but I don't want to lose them either, but I haven't got the I'm not very good at holding that in my head, so I just delegate it to a workflow and it nudges nudges for me.
00:13:05.840 --> 00:13:07.519
And I do that with my invoicing as well.
00:13:07.679 --> 00:13:22.639
I'm not very good at invoicing and reminding people, so I have you know, uh effectively an employee on NA10 who does it, just watches the database and sends out nudges and makes sure everything's ticking over.
00:13:22.799 --> 00:13:25.440
So there's just so many possibilities.
00:13:25.600 --> 00:13:31.360
Um and it's it's really fun uh to to see and and to build.
00:13:31.519 --> 00:13:33.360
So definitely.
00:13:33.600 --> 00:13:36.799
I think a couple of examples I can share of what I've what I've built.
00:13:36.960 --> 00:13:46.960
Um the first one uh is a bit niche for anyone who's doing online meetings, but for me, I do a lot of in-person meetings and I can often be uh driving around the country.
00:13:47.120 --> 00:13:56.080
Uh so having a um workflow that whenever a meeting gets booked into my calendar, it checks to see if there's a location in the meeting.
00:13:56.159 --> 00:14:08.159
And if there is, it will automatically use it will use Google's travel API to be able to calculate an average travel time, add a bit of leeway, and then it can add it as a time blocker on either side of my diary so no one can book something.
00:14:08.399 --> 00:14:11.600
Because using something like Calendly, you can do this.
00:14:11.759 --> 00:14:16.720
You can say, Oh, well, I I'm gonna have a two-hour meeting, but I want to automatically add one hour either side.
00:14:16.879 --> 00:14:20.960
But Calendly doesn't know where you're actually going for that appointment.
00:14:21.200 --> 00:14:25.039
It doesn't know if it's online, it doesn't know if it's you know five hours away.
00:14:25.200 --> 00:14:36.320
Uh so having my workflow do that, that was a huge thing of just making sure that um my calendarly is actually accurate for clients to be able to book, because then when they do book something, it will block out the appropriate travel time.
00:14:36.480 --> 00:14:47.440
I think the favorite one that I've mentioned to financial planners, though, um, and this probably I'd be interested to see what you think about this because it comes down to size of the business.
00:14:47.679 --> 00:14:54.720
But certainly for smaller businesses like mine that want to uh still have that real professional feel to them.
00:14:54.799 --> 00:14:59.200
Uh a lot of mine, I built an entire new client journey workflow.
00:14:59.279 --> 00:15:11.759
So what we would what I was doing there is um when a client books their first meeting, they automatically get a full HTML um agenda for the meeting so they know exactly what to expect.
00:15:11.919 --> 00:15:17.600
It's got loads of content in there that they can click through to on the website, they can download my client agreement, things like that.
00:15:17.919 --> 00:15:30.639
And then uh they also get a text message that comes the day before the meeting to remind them the text message comes from Arboretum, so they get a nice branded text message to say, you know, you're you're don't forget your meeting tomorrow.
00:15:30.799 --> 00:15:42.480
Uh but then also on the other end, because NAN has that information around when the actual meeting is going to take place, it can then trigger stuff for after the meeting.
00:15:42.639 --> 00:15:50.159
So I have like uh follow-up emails that are generated and sent put in my drafts uh from the meeting notes like you described.
00:15:50.320 --> 00:16:00.720
I've also got uh tasks that generate on in my Google tasks certain follow-up activities that I need to do, it automatically works out the deadline based on the meeting.
00:16:00.879 --> 00:16:06.240
And just that keeps everything in my mind because, like you said, it's so impossible to remember all of this.
00:16:06.320 --> 00:16:12.080
Uh, but if you have a streamlined process and you know you want to follow it and it ends perfect for that.
00:16:12.320 --> 00:16:15.840
What would you say then about about the size and size and firms?
00:16:16.320 --> 00:16:20.240
Do you do you see this as more of a small firm thing or a big firm thing?
00:16:20.480 --> 00:16:23.440
Do you think there are any firms even that it's not right for?
00:16:24.720 --> 00:16:38.639
Yeah, I think well, you know, you're a case in point and others that I know as well, the sort of smaller businesses, the one-man bands, are often who've got maybe the the motivation to learn this.
00:16:38.799 --> 00:16:40.080
They're interested in tech.
00:16:40.399 --> 00:16:45.600
That's something that they can get significant way down on their own.
00:16:45.840 --> 00:16:49.440
Um, and you know, it's a very democratised technology.
00:16:49.759 --> 00:16:53.759
This this AI-powered world we now live in is is wonderful.
00:16:53.919 --> 00:17:02.480
Uh the trade-off is time and uh complexity and um troubleshooting because these things do break sometimes.
00:17:02.639 --> 00:17:09.759
So and you know, you can't completely rely on AI to tell you what to do, because often the advice it gives you is out of date.
00:17:09.920 --> 00:17:16.799
It's running off an old interface, and you have to sort of be able to to work it out and do some problem solving.
00:17:17.039 --> 00:17:27.839
So um yeah, I think maybe the sweet spot is between uh maybe a three to even 30 person firm.
00:17:28.000 --> 00:17:35.519
It you know, it's uh brilliant fully advisor who's got a bit of flexibility over their tech stack and back office.
00:17:35.599 --> 00:17:41.839
You know, they're not locked into what they can and can't use so much, they've got a bit of freedom to to do that.
00:17:42.160 --> 00:17:46.480
Um so yeah, I think there's there's a lot of room there.
00:17:46.640 --> 00:17:55.200
Um the the very large firms maybe you know thought they're in in-house um custom software and things, but perhaps not.
00:17:55.440 --> 00:18:02.079
But I think it's perfect for the typical independent independent financial advisor in particular.
00:18:02.400 --> 00:18:05.119
And you know, totally agree, yeah, absolutely.
00:18:05.680 --> 00:18:11.839
So it's when it gets well you say, when it gets to that sort of size, then maybe you're looking at in-house stuff.
00:18:11.920 --> 00:18:18.400
But even, you know, like you say, like very small businesses, this is typically something that they will do a lot of themselves.
00:18:18.640 --> 00:18:29.759
If the medium sized, this is where they probably want someone in maintaining it, someone like yourself who's who's gonna just look after everything for them while it's uh while it's running and make sure there's nothing that goes wrong.
00:18:29.920 --> 00:18:34.960
Um, but the great thing is is that for both of those, it's still a really, really cost-effective solution, right?
00:18:35.039 --> 00:18:42.480
Like it's it's very, very, very easy to maintain because it's so visual, you can identify where stuff's going wrong quite quickly.
00:18:42.720 --> 00:18:49.039
Um so you know, from a perspective of someone needing maintenance on it, it's still pretty low cost, right?
00:18:49.359 --> 00:18:51.039
Yeah, it's very manageable.
00:18:51.200 --> 00:18:59.440
And um, yeah, it's just is I mean, it's it's obviously not error-free.
00:18:59.519 --> 00:19:02.079
Yeah, and like you just said, it it can break.
00:19:02.319 --> 00:19:06.079
But again, you can set up workflows that notify you if there's an error.
00:19:06.240 --> 00:19:10.960
So one of my best workflows in NA10 is a an error handling workflow.
00:19:11.039 --> 00:19:22.000
It will tell it'll send me an email if something's not working, and it will give me the link to the workflow and show me the node where it's broken, and so I can very quickly fix it.
00:19:22.240 --> 00:19:29.599
Um so if you've if you're very if you're keen to do to do this yourself as an advisor, I'd really encourage it.
00:19:29.839 --> 00:19:33.039
It's it's not as intimidating as it perhaps looks.
00:19:33.200 --> 00:19:34.880
You don't need to be a developer.
00:19:35.200 --> 00:19:52.720
Um and like you said earlier, you can get Claude and other AI models to to hold your hand, and certainly at the beginning of learning this, um, so that you can set up some workflows that make a meaningful difference quite early on.
00:19:52.880 --> 00:20:14.240
Um I can't tell you how much time these workflows have saved me with with invoice handling and generation, with proposal writing, and as well the quality is so much better than than it was two years ago had I been doing it all myself, uh, because I've got pre-loaded in there how I would say it and present it on my best day.
00:20:14.480 --> 00:20:23.519
And so it's it's just it's it's fantastic, and you can uh scale so much better with with systems like this.
00:20:23.759 --> 00:20:34.880
Um I think it's got huge potential for financial advice in particular, which has got a lot of administrative overhead and and processes that often hold it back.
00:20:35.680 --> 00:20:36.160
Yeah.
00:20:36.480 --> 00:20:45.839
Um another thing that NAN's got going for it is it's a lot better than something like Zapia at uh its integration of AI, right?
00:20:45.920 --> 00:20:51.200
Like you can you can build in, you mentioned before, you can pull in any kind of uh AI models.
00:20:51.279 --> 00:20:58.799
And I know you've done some work like building things like chatbots, because that that is a thing that you can do in NAN, is you can physically build your own chatbot.
00:20:58.960 --> 00:21:03.599
Have you got any ideas of some use cases that financial planning firms could could use the chatbots for?
00:21:03.680 --> 00:21:13.359
But whether that's front of office or actually back office, weirdly people think of a of a chatbot only as front office, but I can certainly think of some good back office uses.
00:21:14.160 --> 00:21:23.119
Yeah, in fact, the one I'm most excited about comes off some news that just arrived from the FCA, which has just opened its API.
00:21:23.519 --> 00:21:24.640
I saw that, yeah.
00:21:26.480 --> 00:21:39.519
So I've been having a great time with that and um managed to build an internal chat bot which queries the FCA handbook from within NA10.
00:21:39.839 --> 00:21:48.319
So of the really polished version of this would be I'd have a a really nice user interface where I can chat um through that.
00:21:48.640 --> 00:21:58.000
But within the NA10 itself, there's a chat interface and I can just ask it uh a question about uh, say, conflict of interest or something.
00:21:58.160 --> 00:21:59.680
What does the handbook say about that?
00:22:00.160 --> 00:22:06.799
And it will call the API, the relevant section, and then it will talk to an AI model, which then gives me an answer.
00:22:07.039 --> 00:22:22.319
So instead of just searching the wider web and hoping it finds the best sort of open source information on that, it's going directly to the source and feeding that into the AI model and giving me the answer.
00:22:22.480 --> 00:22:26.000
So I see huge potential opportunities there.
00:22:26.640 --> 00:22:36.400
If you've got a new recruit, for instance, and you're trying to train them up and they've got lots of questions, or you're an advisor, you're trying to remember.
00:22:36.559 --> 00:22:37.920
I mean, it's a huge book, isn't it?
00:22:38.079 --> 00:22:40.160
You're trying to remember something in it in there.
00:22:40.400 --> 00:22:42.319
You don't have to manually scour through it.
00:22:42.480 --> 00:22:47.680
You could you could use an internal chatbot to help you navigate it and find find the answer.
00:22:48.000 --> 00:22:51.440
Um that's where I'm most excited, I think, is in the back office.
00:22:51.599 --> 00:22:56.960
Obviously, front office is a bit more, you have to be more careful because of compliance.
00:22:57.119 --> 00:23:10.960
I personally wouldn't be comfortable just putting a a chatbot on an advisor website and linking that to an AI model because does raise questions about who's responsible if that chatbot then says something it shouldn't.
00:23:11.200 --> 00:23:14.799
But there are ways you could control that to a degree.
00:23:15.039 --> 00:23:25.440
So you could have NA10 orchestrate a repository of pre-written answers signed off by compliance on different topics.
00:23:25.599 --> 00:23:41.200
So when the user query comes in from the front end, all the AI does in the back end in NA10 is it discerns what type of query this is and routes it to the relevant repository and then brings the pre-written answer to the front.
00:23:41.359 --> 00:23:45.680
And the fallback would be I don't know, um, I can't answer that.
00:23:45.839 --> 00:23:47.119
Please talk to a member of the team.
00:23:47.279 --> 00:23:48.480
This is how you do it.
00:23:48.640 --> 00:23:52.960
Um so it's a bit I'm a bit more conservative, I think, on the front office.
00:23:53.119 --> 00:23:57.440
Um, but back office, I think there's there's massive potential there.
00:23:57.599 --> 00:23:57.759
Yeah.
00:23:58.000 --> 00:23:59.359
So completely agree.
00:23:59.440 --> 00:24:13.839
I I've not I've not done anything on the front, and it's just I see too many horror stories at the minute of like people using McDonald's chatbot to actually use as a generic AI tool, uh, and they can like manage to get it to to defy all of its previous instructions, things like that.
00:24:13.920 --> 00:24:14.960
So I wouldn't be doing it there.
00:24:15.039 --> 00:24:27.359
Uh one thing that I have seen though is like uh I know a lot of certainly medium-sized businesses, they're using like Slack as an internal um messaging between um between staff.
00:24:27.599 --> 00:24:33.759
And using NAN, you can actually get the language model to sit behind a Slack interface.
00:24:33.920 --> 00:24:48.880
So actually, rather than having to open up a specific um LLM model, you can um you can then just open up your Slack, and it's just like a separate chat that just sits with you know the company, company bot.
00:24:49.119 --> 00:24:52.640
Um and this can replace like I know like Gemini is quite good at that.
00:24:52.720 --> 00:24:58.240
If you're in Google Workspace, you can actually just just talk to Gemini about files and things like that.
00:24:58.400 --> 00:25:04.480
But this is essentially a way that you could replicate that in something like Slack or or something similar, right?
00:25:04.559 --> 00:25:21.839
Is that's I think that's quite a good use from a from an internal perspective of a of a chatbot because it can it can still pull from the same data, it can pull from emails, it can pull from uh your your Google Drive or your Microsoft space, um, and it can still answer questions based off the back of that.
00:25:21.920 --> 00:25:32.400
So you could have, in theory, a chatbot where you say, you know, uh, can you check my calendar for the next few days and tell me which meetings I need to prepare for?
00:25:32.559 --> 00:25:37.839
Uh and then once it gives you that, you could then ask it a follow-up to say, right, can you prep for those meetings for me?
00:25:38.079 --> 00:25:43.680
As long as you'd obviously previously given it instructions of what prep you wanted it to do, that would all be possible, right?
00:25:44.000 --> 00:25:44.559
Absolutely.
00:25:44.720 --> 00:25:45.839
That's that's a great point.
00:25:45.920 --> 00:25:57.039
And I love that about NA10 is you don't one of the friction points always with technology is I've got to learn this new interface, I've got to do I've got to try and wrap my head around this and that.
00:25:57.119 --> 00:26:12.079
And actually, if your business is baked into something like Slack, that's how you communicate, then you can absolutely, like you said, link that up to an NA10 workflow, which then goes to your knowledge base and can you can just talk to it from Slack.
00:26:12.240 --> 00:26:17.680
You don't even have to leave the the discussion area where you're where you're talking to the rest of the team.
00:26:17.839 --> 00:26:32.000
Um and that can might sound like a small thing, but actually, you know, for productivity and um it's focus and concentration, it's it's very useful and that sort of flexibility.
00:26:33.200 --> 00:26:37.759
You um you alluded to this earlier, uh, although I don't think you named it outright.
00:26:37.839 --> 00:26:43.759
Um, because I know you're a big fan of building in open router links within NA10.
00:26:44.000 --> 00:26:52.799
Um do you want to just explain like what open router is and why why you're a big advocate for it in when you're building AI-driven workflows?
00:26:53.839 --> 00:27:05.359
Yeah, I like I like Open Router because uh it essentially what it is is it's a node that points to different AI models depending on the nature of the task.
00:27:05.599 --> 00:27:10.640
So you don't necessarily need a sledgehammer for every task.
00:27:10.720 --> 00:27:14.559
Um quite often a small tool will do the job just as well.
00:27:14.720 --> 00:27:18.559
It might do it quicker, and it will certainly, almost certainly do it cheaper.
00:27:18.799 --> 00:27:37.039
So what's very nice about it is it's one sign up, one login to a platform called Open Router, and then within that it diverts the query or the request or the task to the uh the model that would be best for that particular task.
00:27:37.279 --> 00:27:39.039
So it's it's very useful.
00:27:39.119 --> 00:27:43.759
It set saves you from having to set up multiple AI models manually.
00:27:44.160 --> 00:27:46.480
But you know, there's there's balance there.
00:27:46.640 --> 00:27:54.000
I do also quite like in some instances defining which models I want to talk to for specific tasks.
00:27:54.319 --> 00:27:57.200
Um that might be for data residency reasons.
00:27:57.359 --> 00:27:59.839
You know, I want to keep the request within the UK.
00:28:00.000 --> 00:28:02.799
I don't want it going to a data center overseas.
00:28:02.960 --> 00:28:05.440
So I don't want to delegate that decision to open router.
00:28:05.759 --> 00:28:10.640
I want to have my own control over that particular step.
00:28:10.880 --> 00:28:19.599
But there are there are cases where it's it's you know it is uncomfortable and it's sufficient to delegate that to the to the router itself.
00:28:19.839 --> 00:28:21.839
So it's it's very useful.
00:28:22.960 --> 00:28:32.720
Yeah, you've one of the things that you've mentioned to me before as well is uh you know just the idea that um you know we're in a bit of an AI arms race, like the fair to say.
00:28:32.960 --> 00:28:39.680
And I think most people would agree that Claude is currently winning, but it doesn't mean that's always going to be the case.
00:28:39.839 --> 00:28:46.400
And I did the migration from ChatGBT to Claude earlier this year, it wasn't great.
00:28:46.559 --> 00:28:48.079
Like it's pretty annoying.
00:28:48.160 --> 00:28:49.599
It's a huge friction point.
00:28:49.759 --> 00:28:58.319
So actually, you know, you've mentioned before, by by building in a node where you can basically just a flick of a switch change that model.
00:28:58.880 --> 00:29:08.000
If if Gemini comes through the ranks and all of a sudden next year it's outperforming on every single model compared to Claude, you don't have to migrate all of your stuff over.
00:29:08.079 --> 00:29:09.440
Your workflow is entirely built.
00:29:09.519 --> 00:29:18.319
All you've got to do is just jump in, select the node, and just flick it over and say, actually, for these, for this particular workflow, I now want to use Gemini's layers model, right?
00:29:18.720 --> 00:29:19.839
Absolutely.
00:29:20.160 --> 00:29:22.319
I think, yeah, I think that's the right thinking.
00:29:22.559 --> 00:29:31.039
Um I I love AI, I'm I'm all in on it, but I'm also I I try and think long term about this.
00:29:31.119 --> 00:29:38.079
And you know, we're in a still in an exuberance phase, um, you know, with markets and valuations and all these things.
00:29:38.240 --> 00:29:50.720
And there will come a point, I'm sure, where hard questions will be asked and uh about profit and all the rest of it, and that could that who knows what could happen even in a year or two.
00:29:50.880 --> 00:30:05.200
Um the other thing is it's important to remember that in ten years we'll look back and think how annoying and slow and primitive Claude Code and uh the others are, as much as we're amazed by them now.
00:30:05.440 --> 00:30:19.119
So I think the businesses that will do well, the the practices that will do well long term are those that have as much ownership as possible um over their operational and AI-powered infrastructure.
00:30:19.279 --> 00:30:31.200
And so AI agnosticism is very powerful, data sovereignty is incredibly powerful, and increasingly regulators are looking at operational resilience.
00:30:31.440 --> 00:30:38.640
We had uh the Bank of England, I believe, recently classify some of the big cloud providers as critical third parties.
00:30:39.119 --> 00:30:44.160
Um so there's there's we've got to think strategically.
00:30:44.319 --> 00:30:55.039
These are these aren't just fun tools, they're great tools, but you know, they they could change, they could break, they could be pulled pulled out, like we saw with Fable 5.
00:30:55.920 --> 00:31:01.839
And you know how many come startups were had the rug pulled out of them when that happened, probably quite a few.
00:31:02.000 --> 00:31:14.799
So the great thing about something like NA10 is is it gives you so much more ownership, control, and resilience, and you can pivot very quickly if if things don't go the way you you hope.
00:31:15.200 --> 00:31:19.920
So I think I think this is a good one to look at and and use.
00:31:21.119 --> 00:31:22.160
Absolutely.
00:31:22.799 --> 00:31:24.960
Let's uh we've got to give this a balance.
00:31:25.119 --> 00:31:39.359
Is there any part of NAN that you say it does may maybe falls down, stuff it might not be quite as good at, or even is there any examples in your line of work where you've seen firms try to use it and just get it very wrong?
00:31:41.839 --> 00:31:46.960
Yeah, I think where NA10 falls down is there is a curve a learning curve.
00:31:47.200 --> 00:31:52.960
I think when you start using it seriously, you you're gonna have to invest some time into learning it.
00:31:53.279 --> 00:32:01.039
And with practice, with experience, there comes a point where it becomes a bit second nature.
00:32:01.359 --> 00:32:03.599
But that does take a significant investment.
00:32:03.839 --> 00:32:12.960
So it's it's it's it's on the one hand, user-friendly, beginner-friendly, it's a canvas, it's like painting.
00:32:13.039 --> 00:32:22.799
Anyone can pick up a brush and paint, and but obviously with the same with painting, it's there comes a point where it's quite difficult and you're learning the the mastery of it.
00:32:22.960 --> 00:32:31.599
And um I that's the area where some of the really powerful uh high-leverage workflows live.
00:32:31.680 --> 00:32:34.880
Um, but not all of them, but uh quite a few of them, I would say.
00:32:34.960 --> 00:32:54.079
So if you're trying to join multiple apps together, if you're if you're trying to look at an advisor CRM and how to bring that into the picture, that's going to potentially be quite complex, but of course, a huge potential win because of of the the huge data that's that's in there.
00:32:54.319 --> 00:33:04.720
Uh so I would say yes, that's probably the main drawback uh is the learning curve once you hit that intermediate, more advanced level.
00:33:04.960 --> 00:33:21.200
But there's a lot you can do in the earlier stages and you know build schedulers and um document uh generators, uh proposal writers, all the all these amazing tools that can really save hours of time.
00:33:21.519 --> 00:33:27.759
You you can you can really get some powerful workflows quite early on.
00:33:27.920 --> 00:33:30.559
Uh so I just I would I would just say to be mindful of that.
00:33:30.720 --> 00:33:34.400
It's not going to be a matter of oh, this is always going to be easy.
00:33:34.720 --> 00:33:36.559
But but it is very, very powerful.
00:33:36.640 --> 00:33:39.599
And I would encourage everyone to give it a try.
00:33:40.000 --> 00:33:46.160
Yeah, on that note, actually, that's it's a it's a good point that Phil, because this is probably my favorite question about tech.
00:33:46.400 --> 00:34:00.799
Is obviously we're having this discussion, there will be some degree of someone that that's listening to this and thinking, I mean, I like the sound of it, but the idea of you know you guys describing AI workflows, etc., that terrifies me.
00:34:01.039 --> 00:34:20.639
Um what would be like a just if someone's listening and they've they've maybe just plucked up a little bit of courage to say, I'm gonna give this a go, what would you suggest as just a starter for 10 that they could give the, they could, they could build in an AN that might actually be useful within their business and give them that kind of immediate feedback to go, I should probably invest more time with this?
00:34:21.840 --> 00:34:28.320
Yeah, I would say try, it's gonna sound intimidating, but try as a suitability report generator.
00:34:29.119 --> 00:34:43.039
The biggest hurdle you'll find is if you use something like Google Drive or Microsoft is harder, but if you've got Google Drive, setting up your account with NA10 so that they can talk to each other is the tricky part.
00:34:43.119 --> 00:34:58.639
So if you can get over that, if you can build a folder where you put in some documents, you can then uh maybe I would I would say anonymize this, make sure that you're not we do this with synthetic data before you do anything.
00:34:58.880 --> 00:35:14.159
Um from there, shed have a scheduler at the beginning so it runs every day to check the folder, and in that folder it would just feed the documents in to the AI agent there.
00:35:14.320 --> 00:35:18.079
And all of this you can find by searching in the in the bar on the right.
00:35:18.159 --> 00:35:19.679
You just look for the node you want.
00:35:19.840 --> 00:35:26.559
So you just join all this together and then the output, you could have a first draft suitability report on your desk in the morning.
00:35:26.800 --> 00:35:31.360
All you had to do was put the files in the folder the night before.
00:35:31.599 --> 00:35:36.480
Um the tricky bit, as I said, will be sort of getting your credentials set up.
00:35:36.800 --> 00:35:39.360
But this is where AI is your best friend.
00:35:39.519 --> 00:35:49.039
You just talk to something like Claude or Chat GPT and say, I want to connect NA10 to my Google account, how do I do it?
00:35:49.280 --> 00:35:53.760
And it will take you on the first time, probably maybe 30 minutes, 45 minutes.
00:35:54.000 --> 00:35:56.480
They don't make it easy, but once it's done, it's done.
00:35:56.639 --> 00:36:08.320
And from there, you just drag and drop the nodes on, and you can have yourself a pretty decent suitability report generator right there, and you've probably saved yourself a bit of money on a on a subscription.
00:36:08.480 --> 00:36:11.119
Uh so I would say give that a try.
00:36:11.280 --> 00:36:29.280
And uh if that intimidates you, maybe just try a manual trigger with something more basic, um, you know, like a Google Google Sheet or an NA10 table, see if you can get it to populate another node just with some basic data.
00:36:29.440 --> 00:36:40.000
And then you just you just so you can see how data moves from one part of the canvas to the other, and then just just gradually build from there, build your confidence from there.
00:36:40.239 --> 00:36:46.480
You'll be amazed at what you can do with with no development experience or coding expertise.
00:36:48.639 --> 00:36:49.280
Absolutely.
00:36:49.440 --> 00:36:50.960
That's a that's some great examples.
00:36:51.039 --> 00:36:52.880
I don't think I've got any that I can add to that.
00:36:52.960 --> 00:36:56.480
Um I do think just you know, start with the basics, just give it give it a go.
00:36:56.639 --> 00:37:02.800
I guess maybe the only one would be um, you know, you can look at like following on from meetings, you can have it drafting emails.
00:37:02.880 --> 00:37:08.800
That there is only a couple of nodes to link there, so that can be a little bit user-friendly to give that a go.
00:37:08.960 --> 00:37:14.719
Um or you you mentioned about the the recordings, uh, transcripts, that's also possible.
00:37:14.800 --> 00:37:19.679
You don't have to bring chatbots into the equation, you can just have that triggering off the back of meetings.
00:37:19.840 --> 00:37:22.639
So uh maybe ones like that to to try.
00:37:22.800 --> 00:37:25.360
Uh, but I think I think that's uh I think that's as good.
00:37:25.599 --> 00:37:27.039
We're uh we're done for today.
00:37:27.119 --> 00:37:29.199
So thank you very much for joining me, Phil.
00:37:29.440 --> 00:37:31.119
Uh thank you to everyone listening.
00:37:31.280 --> 00:37:37.119
If you found this one useful, please do leave as a review, like, share, subscribe, all the good stuff.
00:37:37.280 --> 00:37:46.000
Uh don't forget though that if you've got an idea for a piece of tech that you want me to cover, head over to bionicadvisor.co.uk and submit the contact form, and I will be in touch.
00:37:46.159 --> 00:37:49.119
But until next time, look after yourselves, everyone.
00:00:01.679 --> 00:00:09.759
Welcome to another episode of Bionic Advisor, where we take a deep dive into all the amazing pieces of tech being used by financial planning firms up and down the country.
00:00:10.000 --> 00:00:14.880
If you're a financial planner struggling to keep up on the tech treadmill, this is your home.
00:00:15.119 --> 00:00:20.640
For this episode, I'm joined by the amazing Phil Teal, who runs his own business, Teal Pen and Pixel.
00:00:20.719 --> 00:00:21.920
Thanks for joining us, Phil.
00:00:22.399 --> 00:00:23.760
Pleasure to be here.
00:00:24.879 --> 00:00:38.000
So I've known Phil for some time now, and the two of us initially bonded over a mutual love of a tool that I would say is one of the best kept secrets from financial planners.
00:00:38.320 --> 00:00:40.640
And that tool is called NA10.
00:00:41.039 --> 00:00:47.280
I've only met a handful of financial planners who have even heard of NAN, let alone use it day to day.
00:00:47.600 --> 00:00:52.640
But uh for me personally, I'm at the point now where I could not run my business without it.
00:00:52.880 --> 00:00:59.920
So I asked Phil to join me for an episode to have a chat about NA10 because he is a daily NAN user.
00:01:00.079 --> 00:01:05.519
He is building workflows within NAN on a daily basis for financial planning firms.
00:01:05.840 --> 00:01:13.359
And it's such an amazing piece of kit that I honestly think this might be the most useful episode of Bionic Advisor that we ever do.
00:01:13.680 --> 00:01:16.000
And it's just the start, so it's all downhill from here.
00:01:16.079 --> 00:01:17.439
There's nothing gonna be better than this.
00:01:17.840 --> 00:01:20.799
Um let's get into it, shall we?
00:01:21.359 --> 00:01:22.480
Let's do it.
00:01:22.799 --> 00:01:23.120
Okay.
00:01:23.519 --> 00:01:27.359
Okay, so I think first off, nice and easy.
00:01:27.599 --> 00:01:35.760
In plain English, do you just want to explain what NAN is and how you would describe it to a planner who's never touched anything like it?
00:01:36.719 --> 00:01:37.359
Yes.
00:01:37.599 --> 00:01:50.480
Um so NA10 is it's a bit like a canvas where you can drag different apps onto this canvas or workbench and connect them together.
00:01:50.640 --> 00:02:00.719
Um so the idea is you're trying to get different systems talking to each other that might not normally do that, and it's incredibly flexible and versatile.
00:02:00.959 --> 00:02:12.080
So it's going beyond um, you know, for instance, connecting something like Airtable to Google Docs, although you know that's a great type of integration.
00:02:12.560 --> 00:02:28.240
But you can connect um to various platforms and CRMs um that might seem impenetrable through, for instance, something called an HTTP request, which is a special node in NA10.
00:02:28.319 --> 00:02:30.240
Um, and it's just very powerful.
00:02:30.400 --> 00:02:32.159
And um I love it.
00:02:32.240 --> 00:02:39.840
I use it every day, and it powers a lot of my own business operations, and it's incredibly useful for advisors as well.
00:02:40.000 --> 00:02:45.039
And uh I wish more knew about it, so I'm very glad to be here to to talk about it.
00:02:45.759 --> 00:02:46.560
Definitely.
00:02:46.719 --> 00:02:57.520
Um so on um on a previous episode with Dan Graham, I uh talked about Zapia, and there's a lot of similarities, it's fair to say, between Zapia and NAN.
00:02:57.840 --> 00:03:01.919
But uh another one that I hear people talk about quite often is make.
00:03:02.560 --> 00:03:07.919
But how would you say that NAN differs from something like Zapia or Make?
00:03:09.199 --> 00:03:16.719
Yeah, so I would say actually that something like Zapia is great to get started with something like NA10.
00:03:16.879 --> 00:03:21.199
You can go straight to NA10, but Zapia is very beginner-friendly.
00:03:21.520 --> 00:03:27.280
But where NA10 really shines is on quite a few different fronts.
00:03:27.439 --> 00:03:28.960
One is uh cost.
00:03:29.360 --> 00:03:40.080
So once you start scaling up with NA10, um with Zapia, sorry, you find that it can get quite expensive because you're paying per execution.
00:03:40.240 --> 00:03:43.199
Uh, with NA10, you can self-host it.
00:03:43.360 --> 00:03:59.120
So you can put it on uh a computer that you're renting somewhere, something called a VPS, and that could be as little as eight pounds a month, and you can run it as much as you like, really within the limits of that computer.
00:03:59.680 --> 00:04:06.719
Um so as the business grows and as you add people, add workflows to it, it stays very lean.
00:04:07.120 --> 00:04:12.000
Um and uh it ties into another thing I really like about it, which is ownership.
00:04:12.159 --> 00:04:20.800
Um, you can use NA10 on the cloud, and that's where I would recommend most people start as they go to the website and just give it a free trial.
00:04:21.120 --> 00:04:31.279
Um, but when you do put it on your own rented computer, a VPS, then you are effectively it's yours because it's open source.
00:04:31.439 --> 00:04:40.160
And in fact, you can run it on your local machine if you absolutely wanted to, it's a bit more complicated to set up, but um you could do that.
00:04:40.319 --> 00:04:50.959
So um and I I would just say as well that the functionality it offers is just it's it's m far more flexible than uh Zapia.
00:04:51.279 --> 00:04:56.639
Uh some of the workflows you can build are far more complicated if you need them to be.
00:04:56.800 --> 00:05:00.879
Um so it's just a fantastic tool on so many different fronts.
00:05:01.040 --> 00:05:03.040
Uh so yeah, those are three reasons I'd say.
00:05:03.279 --> 00:05:04.959
I'm glad you touched on the the pricing.
00:05:05.040 --> 00:05:17.279
That's a it's a great point because one of the key things that they differ on is uh you know, when you build uh certainly when you build into like multi-flow stuff in Zapia, they actually calculate each step of that as a zap.
00:05:17.519 --> 00:05:24.079
So if you if you have one thing leading to another, that's one, but then if that second thing leads to a third thing, that's actually another zap.
00:05:24.319 --> 00:05:27.519
And you're you're limited, your pricing is based on how many zaps.
00:05:27.600 --> 00:05:31.600
Whereas NA10, one execution is the whole workflow, right?
00:05:31.680 --> 00:05:52.639
So if you just you can build some absolute behemoth of a workflow, even if you're on the cloud, which I think from memory, the starter on the cloud is uh 20 pounds a month, and I you get like two and a half thousand workflow executions, that's a lot of stuff to be going through for just 20 pounds a month, like 2,500 full workflows from start to finish.
00:05:52.800 --> 00:05:54.399
So that's a massive part of it.
00:05:54.639 --> 00:06:04.079
I think you're right though, for me, it's it's a little bit more flexible, it's a little bit more in the way of just it can cope with anything.
00:06:04.240 --> 00:06:11.680
I think with Zapia, you're essentially limited by what connections they've built themselves, you know.
00:06:11.759 --> 00:06:23.439
So it's if if Zapia have decided that the link between Airtable and and Gmail needs to have these triggers and these results, um that's basically all you're gonna get.
00:06:23.680 --> 00:06:36.480
Whereas really, when you get into NA10, because of how powerful the uh the actual coding is, because you you can you can get the code uh code nodes, can't you, in NA10?
00:06:36.959 --> 00:06:44.160
And that can then interpret stuff in a way that uh you would otherwise be reliant on something like Zapia to build.
00:06:44.240 --> 00:06:54.000
Whereas in NA10, you're you're much closer to the full code inside, fair to say, than like something like Zapia, which is just a little bit more plug-and-play.
00:06:54.959 --> 00:06:55.680
Absolutely.
00:06:55.920 --> 00:07:02.319
Yeah, um it you could it's very uh there's a big spectrum of users of NA10.
00:07:02.560 --> 00:07:09.839
So when I started, I was very um unaware of code and JavaScript and all these things.
00:07:10.000 --> 00:07:18.480
And it's actually been a fantastic platform for me to build my confidence in development languages and the sorts of applications I'm building now.
00:07:18.639 --> 00:07:25.839
I I would attribute a lot of it to my time on NA10 because you're learning by uh solving problems.
00:07:26.160 --> 00:07:26.720
By doing it.
00:07:26.800 --> 00:07:27.040
Yeah.
00:07:27.360 --> 00:07:38.959
Trying to figure out how do I get this code node to take this jarble, garbled up mess of something that's come from another app and and streamline it into something that another app can understand.
00:07:39.040 --> 00:07:44.720
And often you need something like a code node to do that, which which is where some of the other platforms maybe are more limited.
00:07:44.959 --> 00:07:47.519
So it is fantastic.
00:07:47.680 --> 00:07:53.120
And the cost as well, you know, obviously clawed code is all the rage right now, and I love clawed code.
00:07:53.199 --> 00:07:58.560
I use it too, but um, you know, if you're pushing that quite hard, it it does get expensive too.
00:07:58.720 --> 00:08:02.879
So I think some of the tiers are something like£100 or£100 a month.
00:08:03.040 --> 00:08:17.360
So the the great thing about NA10 is as you say, it it it's very cost effective, and you can actually talk to AI models from NA10, and you pay as you go as you do it, um, which is actually incredibly cheap.
00:08:17.519 --> 00:08:29.040
I mean, some models are costlier than others, but um you know, you're talking pennies, um, fractions of pennies for some of the calls, the API calls to these different models.
00:08:29.279 --> 00:08:31.519
And the other thing is you can call different ones.
00:08:31.680 --> 00:08:41.039
So you're not just limited to Claude, you can call open AI, perplexity, um you know, um uh all sorts, mistral.
00:08:41.600 --> 00:08:45.360
Just it's for it, you've got some sort of backups built into it as well.
00:08:45.519 --> 00:08:51.120
If one of the providers isn't working when the workflow runs, you can you can fall back to another model.
00:08:51.440 --> 00:08:53.759
So I really like it for that as well.
00:08:54.639 --> 00:08:56.639
Also the other way around as well, though, isn't it?
00:08:56.720 --> 00:09:04.000
Because I think that's a thing that puts people off when they're hearing about we're saying this this is a bit of kit that's very close to full coding.
00:09:04.240 --> 00:09:06.879
And people are like, whoa, whoa, whoa, I'm not a coder.
00:09:06.960 --> 00:09:11.440
Like the thing is you can actually get Clawed to build stuff in NAN for you.
00:09:11.600 --> 00:09:22.720
But I think, and you mentioned Clawed Code, because I do use Claude code, but the thing I still like about NAN above Claude Code is it's so visual, so it's very easy to see the flow of what's actually going on.
00:09:22.799 --> 00:09:24.960
Whereas clawed code, that's pretty much hidden.
00:09:25.039 --> 00:09:28.480
There's no real friendly interface for you to be able to see.
00:09:28.720 --> 00:09:35.279
This is going to trigger this to happen, it's gonna go down this condition path, and then it's gonna result in these potential outcomes.
00:09:35.440 --> 00:09:37.519
You can't really get that in in clawed code.
00:09:37.679 --> 00:09:42.639
So you get the best of both worlds because you can have N8N and you can still have Claude build your workflow in NAN for you.
00:09:43.519 --> 00:09:44.639
That's absolutely right.
00:09:44.720 --> 00:09:45.360
I I agree.
00:09:45.440 --> 00:09:48.559
I'm I'm quite a visual person as well, so I love it for that.
00:09:48.799 --> 00:09:55.679
And yeah, it can be quite intimidating going to Claude Code or something like that's got a command line type interface.
00:09:55.840 --> 00:09:58.080
I think a lot of people get intimidated by that.
00:09:58.240 --> 00:10:07.360
And NA10 is a lot more, you know, you just arrive at the canvas, it's a blank canvas, you have a manual trigger to start, and then just connect something to it.
00:10:07.600 --> 00:10:16.799
And you've got a workflow um as simple as that, and just build on it and see see where your imagination goes with it.
00:10:16.960 --> 00:10:18.879
Um so absolutely.
00:10:19.120 --> 00:10:24.080
Well, so um within your business, this is obviously this is what you're doing day in, day out for financial planning firms.
00:10:24.240 --> 00:10:28.320
You are building these types of workflows and and is a huge part of what you do.
00:10:28.879 --> 00:10:36.720
What's like give me some examples of some common workflows that you would build for an advice firm, some like the bread and butter use cases.
00:10:38.639 --> 00:10:44.320
So um one that's a lot of people are asking about at the moment is a meeting note-taker.
00:10:44.559 --> 00:10:52.960
So there are advisor techs that you can sign up for, which will give you a note-taker, but they can be quite costly.
00:10:53.120 --> 00:11:01.519
So I recently learned as well one of them quite significantly put up its prices, and these charge per advisor per month.
00:11:01.759 --> 00:11:06.480
So if you've got three or four advisors in your business, that's going to add up very quickly.
00:11:06.799 --> 00:11:12.799
So one workflow that I've been working on is an NA10 meeting note taker.
00:11:13.120 --> 00:11:21.919
And the wonderful thing about NA10 is that you can call the services directly that a lot of these advisor techs are actually using under the hood.
00:11:22.080 --> 00:11:23.840
Um I'm not bashing them all.
00:11:23.919 --> 00:11:34.720
I think some of them are quite interesting and useful, but um a lot of them are sort of packaging, something that you could, if you want to, build yourself.
00:11:34.960 --> 00:11:46.720
So you could have it transcribe a meeting for you after listening to the audio, and then summarize it into the structure that you want, which could then go into your CRM.
00:11:46.879 --> 00:11:48.639
So that's one example.
00:11:48.879 --> 00:11:53.360
Uh another example would be a suitability report generator.
00:11:53.919 --> 00:12:23.039
Uh so you feed it uh the fact find, you know, the various meeting notes you've got on the client, put that into a folder, and you have an A10, watch the folder, and when when all the conditions are set, it triggers and creates a brand good suitability report for you, which gets you maybe 70 or 80% of the way there, and then you go over it and check it's right, and then re-ingest it to get it almost to the finish line, and then you polish it off.
00:12:23.279 --> 00:12:24.960
So that's another useful one.
00:12:25.200 --> 00:12:28.879
Uh so and there's just so many uh that's sort of back office.
00:12:28.960 --> 00:12:49.120
I suppose there's also front office too, where um you can use it to join up your website contact form to your mailer light subscriber list, and then you've got a pipeline there to to nudge people uh along um to to book a meeting if they're not ready right there and then.
00:12:49.279 --> 00:12:51.120
So that that's something I use as well.
00:12:51.360 --> 00:13:05.759
If people download a resource from my site, um they might not want to talk straight away, and I understand that, but I don't want to lose them either, but I haven't got the I'm not very good at holding that in my head, so I just delegate it to a workflow and it nudges nudges for me.
00:13:05.840 --> 00:13:07.519
And I do that with my invoicing as well.
00:13:07.679 --> 00:13:22.639
I'm not very good at invoicing and reminding people, so I have you know, uh effectively an employee on NA10 who does it, just watches the database and sends out nudges and makes sure everything's ticking over.
00:13:22.799 --> 00:13:25.440
So there's just so many possibilities.
00:13:25.600 --> 00:13:31.360
Um and it's it's really fun uh to to see and and to build.
00:13:31.519 --> 00:13:33.360
So definitely.
00:13:33.600 --> 00:13:36.799
I think a couple of examples I can share of what I've what I've built.
00:13:36.960 --> 00:13:46.960
Um the first one uh is a bit niche for anyone who's doing online meetings, but for me, I do a lot of in-person meetings and I can often be uh driving around the country.
00:13:47.120 --> 00:13:56.080
Uh so having a um workflow that whenever a meeting gets booked into my calendar, it checks to see if there's a location in the meeting.
00:13:56.159 --> 00:14:08.159
And if there is, it will automatically use it will use Google's travel API to be able to calculate an average travel time, add a bit of leeway, and then it can add it as a time blocker on either side of my diary so no one can book something.
00:14:08.399 --> 00:14:11.600
Because using something like Calendly, you can do this.
00:14:11.759 --> 00:14:16.720
You can say, Oh, well, I I'm gonna have a two-hour meeting, but I want to automatically add one hour either side.
00:14:16.879 --> 00:14:20.960
But Calendly doesn't know where you're actually going for that appointment.
00:14:21.200 --> 00:14:25.039
It doesn't know if it's online, it doesn't know if it's you know five hours away.
00:14:25.200 --> 00:14:36.320
Uh so having my workflow do that, that was a huge thing of just making sure that um my calendarly is actually accurate for clients to be able to book, because then when they do book something, it will block out the appropriate travel time.
00:14:36.480 --> 00:14:47.440
I think the favorite one that I've mentioned to financial planners, though, um, and this probably I'd be interested to see what you think about this because it comes down to size of the business.
00:14:47.679 --> 00:14:54.720
But certainly for smaller businesses like mine that want to uh still have that real professional feel to them.
00:14:54.799 --> 00:14:59.200
Uh a lot of mine, I built an entire new client journey workflow.
00:14:59.279 --> 00:15:11.759
So what we would what I was doing there is um when a client books their first meeting, they automatically get a full HTML um agenda for the meeting so they know exactly what to expect.
00:15:11.919 --> 00:15:17.600
It's got loads of content in there that they can click through to on the website, they can download my client agreement, things like that.
00:15:17.919 --> 00:15:30.639
And then uh they also get a text message that comes the day before the meeting to remind them the text message comes from Arboretum, so they get a nice branded text message to say, you know, you're you're don't forget your meeting tomorrow.
00:15:30.799 --> 00:15:42.480
Uh but then also on the other end, because NAN has that information around when the actual meeting is going to take place, it can then trigger stuff for after the meeting.
00:15:42.639 --> 00:15:50.159
So I have like uh follow-up emails that are generated and sent put in my drafts uh from the meeting notes like you described.
00:15:50.320 --> 00:16:00.720
I've also got uh tasks that generate on in my Google tasks certain follow-up activities that I need to do, it automatically works out the deadline based on the meeting.
00:16:00.879 --> 00:16:06.240
And just that keeps everything in my mind because, like you said, it's so impossible to remember all of this.
00:16:06.320 --> 00:16:12.080
Uh, but if you have a streamlined process and you know you want to follow it and it ends perfect for that.
00:16:12.320 --> 00:16:15.840
What would you say then about about the size and size and firms?
00:16:16.320 --> 00:16:20.240
Do you do you see this as more of a small firm thing or a big firm thing?
00:16:20.480 --> 00:16:23.440
Do you think there are any firms even that it's not right for?
00:16:24.720 --> 00:16:38.639
Yeah, I think well, you know, you're a case in point and others that I know as well, the sort of smaller businesses, the one-man bands, are often who've got maybe the the motivation to learn this.
00:16:38.799 --> 00:16:40.080
They're interested in tech.
00:16:40.399 --> 00:16:45.600
That's something that they can get significant way down on their own.
00:16:45.840 --> 00:16:49.440
Um, and you know, it's a very democratised technology.
00:16:49.759 --> 00:16:53.759
This this AI-powered world we now live in is is wonderful.
00:16:53.919 --> 00:17:02.480
Uh the trade-off is time and uh complexity and um troubleshooting because these things do break sometimes.
00:17:02.639 --> 00:17:09.759
So and you know, you can't completely rely on AI to tell you what to do, because often the advice it gives you is out of date.
00:17:09.920 --> 00:17:16.799
It's running off an old interface, and you have to sort of be able to to work it out and do some problem solving.
00:17:17.039 --> 00:17:27.839
So um yeah, I think maybe the sweet spot is between uh maybe a three to even 30 person firm.
00:17:28.000 --> 00:17:35.519
It you know, it's uh brilliant fully advisor who's got a bit of flexibility over their tech stack and back office.
00:17:35.599 --> 00:17:41.839
You know, they're not locked into what they can and can't use so much, they've got a bit of freedom to to do that.
00:17:42.160 --> 00:17:46.480
Um so yeah, I think there's there's a lot of room there.
00:17:46.640 --> 00:17:55.200
Um the the very large firms maybe you know thought they're in in-house um custom software and things, but perhaps not.
00:17:55.440 --> 00:18:02.079
But I think it's perfect for the typical independent independent financial advisor in particular.
00:18:02.400 --> 00:18:05.119
And you know, totally agree, yeah, absolutely.
00:18:05.680 --> 00:18:11.839
So it's when it gets well you say, when it gets to that sort of size, then maybe you're looking at in-house stuff.
00:18:11.920 --> 00:18:18.400
But even, you know, like you say, like very small businesses, this is typically something that they will do a lot of themselves.
00:18:18.640 --> 00:18:29.759
If the medium sized, this is where they probably want someone in maintaining it, someone like yourself who's who's gonna just look after everything for them while it's uh while it's running and make sure there's nothing that goes wrong.
00:18:29.920 --> 00:18:34.960
Um, but the great thing is is that for both of those, it's still a really, really cost-effective solution, right?
00:18:35.039 --> 00:18:42.480
Like it's it's very, very, very easy to maintain because it's so visual, you can identify where stuff's going wrong quite quickly.
00:18:42.720 --> 00:18:49.039
Um so you know, from a perspective of someone needing maintenance on it, it's still pretty low cost, right?
00:18:49.359 --> 00:18:51.039
Yeah, it's very manageable.
00:18:51.200 --> 00:18:59.440
And um, yeah, it's just is I mean, it's it's obviously not error-free.
00:18:59.519 --> 00:19:02.079
Yeah, and like you just said, it it can break.
00:19:02.319 --> 00:19:06.079
But again, you can set up workflows that notify you if there's an error.
00:19:06.240 --> 00:19:10.960
So one of my best workflows in NA10 is a an error handling workflow.
00:19:11.039 --> 00:19:22.000
It will tell it'll send me an email if something's not working, and it will give me the link to the workflow and show me the node where it's broken, and so I can very quickly fix it.
00:19:22.240 --> 00:19:29.599
Um so if you've if you're very if you're keen to do to do this yourself as an advisor, I'd really encourage it.
00:19:29.839 --> 00:19:33.039
It's it's not as intimidating as it perhaps looks.
00:19:33.200 --> 00:19:34.880
You don't need to be a developer.
00:19:35.200 --> 00:19:52.720
Um and like you said earlier, you can get Claude and other AI models to to hold your hand, and certainly at the beginning of learning this, um, so that you can set up some workflows that make a meaningful difference quite early on.
00:19:52.880 --> 00:20:14.240
Um I can't tell you how much time these workflows have saved me with with invoice handling and generation, with proposal writing, and as well the quality is so much better than than it was two years ago had I been doing it all myself, uh, because I've got pre-loaded in there how I would say it and present it on my best day.
00:20:14.480 --> 00:20:23.519
And so it's it's just it's it's fantastic, and you can uh scale so much better with with systems like this.
00:20:23.759 --> 00:20:34.880
Um I think it's got huge potential for financial advice in particular, which has got a lot of administrative overhead and and processes that often hold it back.
00:20:35.680 --> 00:20:36.160
Yeah.
00:20:36.480 --> 00:20:45.839
Um another thing that NAN's got going for it is it's a lot better than something like Zapia at uh its integration of AI, right?
00:20:45.920 --> 00:20:51.200
Like you can you can build in, you mentioned before, you can pull in any kind of uh AI models.
00:20:51.279 --> 00:20:58.799
And I know you've done some work like building things like chatbots, because that that is a thing that you can do in NAN, is you can physically build your own chatbot.
00:20:58.960 --> 00:21:03.599
Have you got any ideas of some use cases that financial planning firms could could use the chatbots for?
00:21:03.680 --> 00:21:13.359
But whether that's front of office or actually back office, weirdly people think of a of a chatbot only as front office, but I can certainly think of some good back office uses.
00:21:14.160 --> 00:21:23.119
Yeah, in fact, the one I'm most excited about comes off some news that just arrived from the FCA, which has just opened its API.
00:21:23.519 --> 00:21:24.640
I saw that, yeah.
00:21:26.480 --> 00:21:39.519
So I've been having a great time with that and um managed to build an internal chat bot which queries the FCA handbook from within NA10.
00:21:39.839 --> 00:21:48.319
So of the really polished version of this would be I'd have a a really nice user interface where I can chat um through that.
00:21:48.640 --> 00:21:58.000
But within the NA10 itself, there's a chat interface and I can just ask it uh a question about uh, say, conflict of interest or something.
00:21:58.160 --> 00:21:59.680
What does the handbook say about that?
00:22:00.160 --> 00:22:06.799
And it will call the API, the relevant section, and then it will talk to an AI model, which then gives me an answer.
00:22:07.039 --> 00:22:22.319
So instead of just searching the wider web and hoping it finds the best sort of open source information on that, it's going directly to the source and feeding that into the AI model and giving me the answer.
00:22:22.480 --> 00:22:26.000
So I see huge potential opportunities there.
00:22:26.640 --> 00:22:36.400
If you've got a new recruit, for instance, and you're trying to train them up and they've got lots of questions, or you're an advisor, you're trying to remember.
00:22:36.559 --> 00:22:37.920
I mean, it's a huge book, isn't it?
00:22:38.079 --> 00:22:40.160
You're trying to remember something in it in there.
00:22:40.400 --> 00:22:42.319
You don't have to manually scour through it.
00:22:42.480 --> 00:22:47.680
You could you could use an internal chatbot to help you navigate it and find find the answer.
00:22:48.000 --> 00:22:51.440
Um that's where I'm most excited, I think, is in the back office.
00:22:51.599 --> 00:22:56.960
Obviously, front office is a bit more, you have to be more careful because of compliance.
00:22:57.119 --> 00:23:10.960
I personally wouldn't be comfortable just putting a a chatbot on an advisor website and linking that to an AI model because does raise questions about who's responsible if that chatbot then says something it shouldn't.
00:23:11.200 --> 00:23:14.799
But there are ways you could control that to a degree.
00:23:15.039 --> 00:23:25.440
So you could have NA10 orchestrate a repository of pre-written answers signed off by compliance on different topics.
00:23:25.599 --> 00:23:41.200
So when the user query comes in from the front end, all the AI does in the back end in NA10 is it discerns what type of query this is and routes it to the relevant repository and then brings the pre-written answer to the front.
00:23:41.359 --> 00:23:45.680
And the fallback would be I don't know, um, I can't answer that.
00:23:45.839 --> 00:23:47.119
Please talk to a member of the team.
00:23:47.279 --> 00:23:48.480
This is how you do it.
00:23:48.640 --> 00:23:52.960
Um so it's a bit I'm a bit more conservative, I think, on the front office.
00:23:53.119 --> 00:23:57.440
Um, but back office, I think there's there's massive potential there.
00:23:57.599 --> 00:23:57.759
Yeah.
00:23:58.000 --> 00:23:59.359
So completely agree.
00:23:59.440 --> 00:24:13.839
I I've not I've not done anything on the front, and it's just I see too many horror stories at the minute of like people using McDonald's chatbot to actually use as a generic AI tool, uh, and they can like manage to get it to to defy all of its previous instructions, things like that.
00:24:13.920 --> 00:24:14.960
So I wouldn't be doing it there.
00:24:15.039 --> 00:24:27.359
Uh one thing that I have seen though is like uh I know a lot of certainly medium-sized businesses, they're using like Slack as an internal um messaging between um between staff.
00:24:27.599 --> 00:24:33.759
And using NAN, you can actually get the language model to sit behind a Slack interface.
00:24:33.920 --> 00:24:48.880
So actually, rather than having to open up a specific um LLM model, you can um you can then just open up your Slack, and it's just like a separate chat that just sits with you know the company, company bot.
00:24:49.119 --> 00:24:52.640
Um and this can replace like I know like Gemini is quite good at that.
00:24:52.720 --> 00:24:58.240
If you're in Google Workspace, you can actually just just talk to Gemini about files and things like that.
00:24:58.400 --> 00:25:04.480
But this is essentially a way that you could replicate that in something like Slack or or something similar, right?
00:25:04.559 --> 00:25:21.839
Is that's I think that's quite a good use from a from an internal perspective of a of a chatbot because it can it can still pull from the same data, it can pull from emails, it can pull from uh your your Google Drive or your Microsoft space, um, and it can still answer questions based off the back of that.
00:25:21.920 --> 00:25:32.400
So you could have, in theory, a chatbot where you say, you know, uh, can you check my calendar for the next few days and tell me which meetings I need to prepare for?
00:25:32.559 --> 00:25:37.839
Uh and then once it gives you that, you could then ask it a follow-up to say, right, can you prep for those meetings for me?
00:25:38.079 --> 00:25:43.680
As long as you'd obviously previously given it instructions of what prep you wanted it to do, that would all be possible, right?
00:25:44.000 --> 00:25:44.559
Absolutely.
00:25:44.720 --> 00:25:45.839
That's that's a great point.
00:25:45.920 --> 00:25:57.039
And I love that about NA10 is you don't one of the friction points always with technology is I've got to learn this new interface, I've got to do I've got to try and wrap my head around this and that.
00:25:57.119 --> 00:26:12.079
And actually, if your business is baked into something like Slack, that's how you communicate, then you can absolutely, like you said, link that up to an NA10 workflow, which then goes to your knowledge base and can you can just talk to it from Slack.
00:26:12.240 --> 00:26:17.680
You don't even have to leave the the discussion area where you're where you're talking to the rest of the team.
00:26:17.839 --> 00:26:32.000
Um and that can might sound like a small thing, but actually, you know, for productivity and um it's focus and concentration, it's it's very useful and that sort of flexibility.
00:26:33.200 --> 00:26:37.759
You um you alluded to this earlier, uh, although I don't think you named it outright.
00:26:37.839 --> 00:26:43.759
Um, because I know you're a big fan of building in open router links within NA10.
00:26:44.000 --> 00:26:52.799
Um do you want to just explain like what open router is and why why you're a big advocate for it in when you're building AI-driven workflows?
00:26:53.839 --> 00:27:05.359
Yeah, I like I like Open Router because uh it essentially what it is is it's a node that points to different AI models depending on the nature of the task.
00:27:05.599 --> 00:27:10.640
So you don't necessarily need a sledgehammer for every task.
00:27:10.720 --> 00:27:14.559
Um quite often a small tool will do the job just as well.
00:27:14.720 --> 00:27:18.559
It might do it quicker, and it will certainly, almost certainly do it cheaper.
00:27:18.799 --> 00:27:37.039
So what's very nice about it is it's one sign up, one login to a platform called Open Router, and then within that it diverts the query or the request or the task to the uh the model that would be best for that particular task.
00:27:37.279 --> 00:27:39.039
So it's it's very useful.
00:27:39.119 --> 00:27:43.759
It set saves you from having to set up multiple AI models manually.
00:27:44.160 --> 00:27:46.480
But you know, there's there's balance there.
00:27:46.640 --> 00:27:54.000
I do also quite like in some instances defining which models I want to talk to for specific tasks.
00:27:54.319 --> 00:27:57.200
Um that might be for data residency reasons.
00:27:57.359 --> 00:27:59.839
You know, I want to keep the request within the UK.
00:28:00.000 --> 00:28:02.799
I don't want it going to a data center overseas.
00:28:02.960 --> 00:28:05.440
So I don't want to delegate that decision to open router.
00:28:05.759 --> 00:28:10.640
I want to have my own control over that particular step.
00:28:10.880 --> 00:28:19.599
But there are there are cases where it's it's you know it is uncomfortable and it's sufficient to delegate that to the to the router itself.
00:28:19.839 --> 00:28:21.839
So it's it's very useful.
00:28:22.960 --> 00:28:32.720
Yeah, you've one of the things that you've mentioned to me before as well is uh you know just the idea that um you know we're in a bit of an AI arms race, like the fair to say.
00:28:32.960 --> 00:28:39.680
And I think most people would agree that Claude is currently winning, but it doesn't mean that's always going to be the case.
00:28:39.839 --> 00:28:46.400
And I did the migration from ChatGBT to Claude earlier this year, it wasn't great.
00:28:46.559 --> 00:28:48.079
Like it's pretty annoying.
00:28:48.160 --> 00:28:49.599
It's a huge friction point.
00:28:49.759 --> 00:28:58.319
So actually, you know, you've mentioned before, by by building in a node where you can basically just a flick of a switch change that model.
00:28:58.880 --> 00:29:08.000
If if Gemini comes through the ranks and all of a sudden next year it's outperforming on every single model compared to Claude, you don't have to migrate all of your stuff over.
00:29:08.079 --> 00:29:09.440
Your workflow is entirely built.
00:29:09.519 --> 00:29:18.319
All you've got to do is just jump in, select the node, and just flick it over and say, actually, for these, for this particular workflow, I now want to use Gemini's layers model, right?
00:29:18.720 --> 00:29:19.839
Absolutely.
00:29:20.160 --> 00:29:22.319
I think, yeah, I think that's the right thinking.
00:29:22.559 --> 00:29:31.039
Um I I love AI, I'm I'm all in on it, but I'm also I I try and think long term about this.
00:29:31.119 --> 00:29:38.079
And you know, we're in a still in an exuberance phase, um, you know, with markets and valuations and all these things.
00:29:38.240 --> 00:29:50.720
And there will come a point, I'm sure, where hard questions will be asked and uh about profit and all the rest of it, and that could that who knows what could happen even in a year or two.
00:29:50.880 --> 00:30:05.200
Um the other thing is it's important to remember that in ten years we'll look back and think how annoying and slow and primitive Claude Code and uh the others are, as much as we're amazed by them now.
00:30:05.440 --> 00:30:19.119
So I think the businesses that will do well, the the practices that will do well long term are those that have as much ownership as possible um over their operational and AI-powered infrastructure.
00:30:19.279 --> 00:30:31.200
And so AI agnosticism is very powerful, data sovereignty is incredibly powerful, and increasingly regulators are looking at operational resilience.
00:30:31.440 --> 00:30:38.640
We had uh the Bank of England, I believe, recently classify some of the big cloud providers as critical third parties.
00:30:39.119 --> 00:30:44.160
Um so there's there's we've got to think strategically.
00:30:44.319 --> 00:30:55.039
These are these aren't just fun tools, they're great tools, but you know, they they could change, they could break, they could be pulled pulled out, like we saw with Fable 5.
00:30:55.920 --> 00:31:01.839
And you know how many come startups were had the rug pulled out of them when that happened, probably quite a few.
00:31:02.000 --> 00:31:14.799
So the great thing about something like NA10 is is it gives you so much more ownership, control, and resilience, and you can pivot very quickly if if things don't go the way you you hope.
00:31:15.200 --> 00:31:19.920
So I think I think this is a good one to look at and and use.
00:31:21.119 --> 00:31:22.160
Absolutely.
00:31:22.799 --> 00:31:24.960
Let's uh we've got to give this a balance.
00:31:25.119 --> 00:31:39.359
Is there any part of NAN that you say it does may maybe falls down, stuff it might not be quite as good at, or even is there any examples in your line of work where you've seen firms try to use it and just get it very wrong?
00:31:41.839 --> 00:31:46.960
Yeah, I think where NA10 falls down is there is a curve a learning curve.
00:31:47.200 --> 00:31:52.960
I think when you start using it seriously, you you're gonna have to invest some time into learning it.
00:31:53.279 --> 00:32:01.039
And with practice, with experience, there comes a point where it becomes a bit second nature.
00:32:01.359 --> 00:32:03.599
But that does take a significant investment.
00:32:03.839 --> 00:32:12.960
So it's it's it's it's on the one hand, user-friendly, beginner-friendly, it's a canvas, it's like painting.
00:32:13.039 --> 00:32:22.799
Anyone can pick up a brush and paint, and but obviously with the same with painting, it's there comes a point where it's quite difficult and you're learning the the mastery of it.
00:32:22.960 --> 00:32:31.599
And um I that's the area where some of the really powerful uh high-leverage workflows live.
00:32:31.680 --> 00:32:34.880
Um, but not all of them, but uh quite a few of them, I would say.
00:32:34.960 --> 00:32:54.079
So if you're trying to join multiple apps together, if you're if you're trying to look at an advisor CRM and how to bring that into the picture, that's going to potentially be quite complex, but of course, a huge potential win because of of the the huge data that's that's in there.
00:32:54.319 --> 00:33:04.720
Uh so I would say yes, that's probably the main drawback uh is the learning curve once you hit that intermediate, more advanced level.
00:33:04.960 --> 00:33:21.200
But there's a lot you can do in the earlier stages and you know build schedulers and um document uh generators, uh proposal writers, all the all these amazing tools that can really save hours of time.
00:33:21.519 --> 00:33:27.759
You you can you can really get some powerful workflows quite early on.
00:33:27.920 --> 00:33:30.559
Uh so I just I would I would just say to be mindful of that.
00:33:30.720 --> 00:33:34.400
It's not going to be a matter of oh, this is always going to be easy.
00:33:34.720 --> 00:33:36.559
But but it is very, very powerful.
00:33:36.640 --> 00:33:39.599
And I would encourage everyone to give it a try.
00:33:40.000 --> 00:33:46.160
Yeah, on that note, actually, that's it's a it's a good point that Phil, because this is probably my favorite question about tech.
00:33:46.400 --> 00:34:00.799
Is obviously we're having this discussion, there will be some degree of someone that that's listening to this and thinking, I mean, I like the sound of it, but the idea of you know you guys describing AI workflows, etc., that terrifies me.
00:34:01.039 --> 00:34:20.639
Um what would be like a just if someone's listening and they've they've maybe just plucked up a little bit of courage to say, I'm gonna give this a go, what would you suggest as just a starter for 10 that they could give the, they could, they could build in an AN that might actually be useful within their business and give them that kind of immediate feedback to go, I should probably invest more time with this?
00:34:21.840 --> 00:34:28.320
Yeah, I would say try, it's gonna sound intimidating, but try as a suitability report generator.
00:34:29.119 --> 00:34:43.039
The biggest hurdle you'll find is if you use something like Google Drive or Microsoft is harder, but if you've got Google Drive, setting up your account with NA10 so that they can talk to each other is the tricky part.
00:34:43.119 --> 00:34:58.639
So if you can get over that, if you can build a folder where you put in some documents, you can then uh maybe I would I would say anonymize this, make sure that you're not we do this with synthetic data before you do anything.
00:34:58.880 --> 00:35:14.159
Um from there, shed have a scheduler at the beginning so it runs every day to check the folder, and in that folder it would just feed the documents in to the AI agent there.
00:35:14.320 --> 00:35:18.079
And all of this you can find by searching in the in the bar on the right.
00:35:18.159 --> 00:35:19.679
You just look for the node you want.
00:35:19.840 --> 00:35:26.559
So you just join all this together and then the output, you could have a first draft suitability report on your desk in the morning.
00:35:26.800 --> 00:35:31.360
All you had to do was put the files in the folder the night before.
00:35:31.599 --> 00:35:36.480
Um the tricky bit, as I said, will be sort of getting your credentials set up.
00:35:36.800 --> 00:35:39.360
But this is where AI is your best friend.
00:35:39.519 --> 00:35:49.039
You just talk to something like Claude or Chat GPT and say, I want to connect NA10 to my Google account, how do I do it?
00:35:49.280 --> 00:35:53.760
And it will take you on the first time, probably maybe 30 minutes, 45 minutes.
00:35:54.000 --> 00:35:56.480
They don't make it easy, but once it's done, it's done.
00:35:56.639 --> 00:36:08.320
And from there, you just drag and drop the nodes on, and you can have yourself a pretty decent suitability report generator right there, and you've probably saved yourself a bit of money on a on a subscription.
00:36:08.480 --> 00:36:11.119
Uh so I would say give that a try.
00:36:11.280 --> 00:36:29.280
And uh if that intimidates you, maybe just try a manual trigger with something more basic, um, you know, like a Google Google Sheet or an NA10 table, see if you can get it to populate another node just with some basic data.
00:36:29.440 --> 00:36:40.000
And then you just you just so you can see how data moves from one part of the canvas to the other, and then just just gradually build from there, build your confidence from there.
00:36:40.239 --> 00:36:46.480
You'll be amazed at what you can do with with no development experience or coding expertise.
00:36:48.639 --> 00:36:49.280
Absolutely.
00:36:49.440 --> 00:36:50.960
That's a that's some great examples.
00:36:51.039 --> 00:36:52.880
I don't think I've got any that I can add to that.
00:36:52.960 --> 00:36:56.480
Um I do think just you know, start with the basics, just give it give it a go.
00:36:56.639 --> 00:37:02.800
I guess maybe the only one would be um, you know, you can look at like following on from meetings, you can have it drafting emails.
00:37:02.880 --> 00:37:08.800
That there is only a couple of nodes to link there, so that can be a little bit user-friendly to give that a go.
00:37:08.960 --> 00:37:14.719
Um or you you mentioned about the the recordings, uh, transcripts, that's also possible.
00:37:14.800 --> 00:37:19.679
You don't have to bring chatbots into the equation, you can just have that triggering off the back of meetings.
00:37:19.840 --> 00:37:22.639
So uh maybe ones like that to to try.
00:37:22.800 --> 00:37:25.360
Uh, but I think I think that's uh I think that's as good.
00:37:25.599 --> 00:37:27.039
We're uh we're done for today.
00:37:27.119 --> 00:37:29.199
So thank you very much for joining me, Phil.
00:37:29.440 --> 00:37:31.119
Uh thank you to everyone listening.
00:37:31.280 --> 00:37:37.119
If you found this one useful, please do leave as a review, like, share, subscribe, all the good stuff.
00:37:37.280 --> 00:37:46.000
Uh don't forget though that if you've got an idea for a piece of tech that you want me to cover, head over to bionicadvisor.co.uk and submit the contact form, and I will be in touch.
00:37:46.159 --> 00:37:49.119
But until next time, look after yourselves, everyone.
