TRANSKRIP 🔗
WEBVTT
00:00:00.160 --> 00:00:05.759
How you can actually stop working on prompts, but rather work on what the model already knows about you when you sit down.
00:00:06.000 --> 00:00:21.039
To have an agent who has his mind of his own if it configured correctly, I think literally it can be an employee that helps you to do work while you are away and to remind you about things that you may forget.
00:00:21.280 --> 00:00:25.039
My chief of staff is very straightforward.
00:00:25.359 --> 00:00:38.560
So he will say that it is a bad week for you, you didn't hit your target, your pipeline is zero, um, you don't have any kind of coffee meetings with anybody next week, you need to get one now.
00:00:38.719 --> 00:00:45.439
So we cut it from let's say a 15-minute kind of process to probably like a two-minute kind of process.
00:00:45.759 --> 00:00:50.079
It was just a two-three-day kind of project.
00:00:50.719 --> 00:01:05.280
So the hack that I came up with and I want to share with many people is that so there are a lot of processes within companies that don't have SOP by rambling this for let's say half an hour or so.
00:01:05.439 --> 00:01:17.359
I have the full documentation there, but I think a lot of times we fall into that rabbit hole whereby building infrastructure uh more than doing what we're supposed to do.
00:01:20.560 --> 00:01:24.400
Hi everyone, I'm Eric here, host of the Code Rift Podcast.
00:01:24.480 --> 00:01:27.120
Uh with me we have uh Yao Hong.
00:01:27.599 --> 00:01:28.159
Hello.
00:01:28.400 --> 00:01:31.920
Hi and uh very special guest, uh Benjamin.
00:01:32.000 --> 00:01:33.840
Uh Ben, welcome to the podcast.
00:01:33.920 --> 00:01:34.799
Thanks for joining us.
00:01:35.120 --> 00:01:36.959
Hi, thank you.
00:01:38.319 --> 00:01:51.120
And and this is gonna be a lot more hands-on as compared to our um previous few interviews that we did, which was super nice and very helpful for like the ways that we think about AI and how we how we live with them.
00:01:51.359 --> 00:01:54.319
Um but today we have Ben, Benjamin Huang.
00:01:54.480 --> 00:02:04.879
Um yeah, so Ben's going to kind of walk us through basically how what he's been doing with AI, how he's been thinking about it.
00:02:05.200 --> 00:02:08.800
Ben, you co-founded Omnidesk and you ran it for eight years.
00:02:08.960 --> 00:02:17.120
Uh you're also now the founder of firstpress.ai and you ship a lot with Cloud Code with Hermes.
00:02:17.520 --> 00:02:20.400
Uh yeah, and and welcome, very warm welcome to the show.
00:02:20.719 --> 00:02:22.080
Thanks for having me here.
00:02:22.240 --> 00:02:22.560
Yeah.
00:02:22.800 --> 00:02:31.919
Um but maybe before you even introduce yourself, uh Ben, do you mind just sharing a bit about you know who this episode is for and what might they get out of it today?
00:02:32.719 --> 00:02:54.479
So I think who this is really for, I think it usually means um probably anybody who's actually managing a team or uh which is like probably more
Load full transcript
than three people, or as long as you are someone who is like uh SME owner, heads of marketing and things like that, or anybody who's like running like more than one business, two business.
00:02:55.120 --> 00:02:58.879
So I think this is yeah, quite widespread.
00:02:59.039 --> 00:03:07.680
Uh unless like to the stage whereby uh you have never had to re write a weekly report to anybody else, then probably this won't actually land up with you.
00:03:07.919 --> 00:03:10.159
So that's probably one of the tests.
00:03:10.400 --> 00:03:26.080
Yeah, so what you get out of um watching this, I think is more to understanding what context is in terms of the AI world and how you can actually stop working on prompts, but rather work on what the model already knows about you when you sit down.
00:03:26.319 --> 00:03:43.840
So basically, cloud code, chat GBT, um anything that you start working on, they already know your background, uh what you need to know, and basically come to you with an answer that actually you really want rather than uh depending on prompts.
00:03:44.080 --> 00:03:44.719
Yep.
00:03:45.919 --> 00:03:54.639
Yeah, and I think that's really helpful because um I think all of us are still experimenting, all of us have our own stacks that we are kind of building with.
00:03:54.800 --> 00:04:04.479
Um but I think it's it's always very valuable to have a peek into a practitioner's uh kind of infrastructure and and code bases.
00:04:04.639 --> 00:04:08.719
Uh and and you do have a GitHub repo to share at the end.
00:04:09.599 --> 00:04:12.319
So, Ben, do you mind just sharing a bit about yourself?
00:04:12.560 --> 00:04:13.919
Like who are you?
00:04:14.319 --> 00:04:17.360
What have you been working on and how do you get into AI?
00:04:17.439 --> 00:04:18.079
Yeah.
00:04:18.720 --> 00:04:26.160
Um so I spent eight years as the co-founder and managing director of Omnidask.
00:04:26.240 --> 00:04:28.800
So we sell standing desk if you're not aware.
00:04:29.040 --> 00:04:33.680
Um so in that eight years, so I've been running the business end-to-end.
00:04:34.000 --> 00:04:39.360
So whether is it um making decisions for the products, uh for marketing and things like that.
00:04:39.600 --> 00:04:45.439
So the entire business um uh updates and everything, it falls on me.
00:04:45.600 --> 00:04:48.560
So I think that's what I've been doing for eight years or so.
00:04:48.800 --> 00:04:58.000
So um I think the main change that came about was probably um between the past 12 months or so.
00:04:58.160 --> 00:05:21.439
So I think that's where um yeah, that the AI kind of things actually start to um land on us and uh whether it's um business owners or students or anybody, yeah, they can feel that hey, um AI is definitely gonna make a difference, it's no longer just a bubble that actually just yeah, just come and go and things like that.
00:05:21.680 --> 00:05:32.560
So when I was in OmniDesk, um time is actually precious, but uh given that AI is making such
a tremendous kind of impact, so I did delve myself into it.
00:05:32.720 --> 00:05:38.319
So I think like most of the people uh you start off with Chat GPT, so the easiest of it all.
00:05:38.560 --> 00:05:44.079
So yeah, to key in a prompt and then to uh to get what I want.
00:05:44.240 --> 00:06:04.240
So the most basic one was that um the copy on the website, so to have uh whatever that I have there, and then I tell ChatGPT okay, hey, can we improve this copy to be more um sophisticated or rather uh can be more conversion um conversion kind of angle and things like that.
00:06:04.399 --> 00:06:08.000
So that kind of work, so it's like a copy and paste kind of thing.
00:06:08.240 --> 00:06:13.839
Then uh as we move on, I think um I'm a very data person.
00:06:14.240 --> 00:06:19.439
So when I get my team to give me a report, I look at the data very stringently.
00:06:19.600 --> 00:06:36.399
So because data is part of my decision-making process, so I want accurate data, and anybody can give you data, but the fact is that uh when I want to process data, I have a certain framework that I have in mind whereby not everybody can provide.
00:06:36.639 --> 00:06:40.160
So a lot of times I have to do it manually after they give me an Excel sheet.
00:06:40.240 --> 00:06:44.800
So, okay, this data it needs to be calculated in a different way, and things like that.
00:06:45.040 --> 00:06:50.399
So I think that was one of the main things that uh came to mind when um AI came here.
00:06:50.560 --> 00:06:58.240
So, how can I generate those kind of reports, those kind of numbers that I can understand and make decisions straight away?
00:06:58.480 --> 00:07:00.480
So that was where I tried, okay.
00:07:00.639 --> 00:07:06.800
So, whenever let's say marketing gives me the report for uh Facebook, for Google, and things like that.
00:07:06.959 --> 00:07:10.879
So I actually tried using like ChatGPT to actually analyze those data.
00:07:11.040 --> 00:07:30.560
So to find out, so what is the performance like in actual fact, rather than having agencies tell me, so uh hey, this is doing well, uh, your ROAS is this is this, but having AI there, it was actually like giving me more in-depth kind of um analysis, so the trends and things like that.
00:07:30.720 --> 00:07:37.600
So it started to give me like uh proper kind of data, and that's where I can make proper decisions.
00:07:37.839 --> 00:07:40.480
So then comes to the next level.
00:07:40.800 --> 00:07:43.199
So, how can I automate this?
00:07:43.439 --> 00:07:52.480
Because yeah, for me, yeah, I mean every week they came to come to me, then I have to put it into um chat GPT and then hey, generate the same thing over and over.
00:07:52.560 --> 00:07:55.360
It's like repeating the same process, and I'm still spending a lot of time.
00:07:55.439 --> 00:07:57.040
So, how can I automate that?
00:07:57.199 --> 00:08:08.079
So, I think that's where um yeah, after Chat GPT is like uh I try cloud because people were saying
that uh cloud has more um uh automations available, things like that.
00:08:08.160 --> 00:08:13.120
So, yeah, but when I tried into just double into cloud, it's like isn't it the same?
00:08:13.199 --> 00:08:18.000
Because it's like yeah, the same kind of uh UI and things like that, so not much.
00:08:18.399 --> 00:08:21.199
Uh that's where I go even deeper.
00:08:21.360 --> 00:08:26.720
So that's where cloud code came about, and I think that's where it opens everything up.
00:08:27.120 --> 00:08:33.120
So initially, when cloud code uh came to me, it's like why do I need cloud code?
00:08:33.200 --> 00:08:40.720
Because at the end of the day, it's like I just need a report, I don't need an app, I don't need let's say uh a website.
00:08:40.960 --> 00:08:45.200
So why am I going into the the CLI and things like that?
00:08:45.279 --> 00:08:47.039
So what's the necessity there?
00:08:47.519 --> 00:08:50.879
And I was curious and I just double into it.
00:08:50.960 --> 00:09:03.360
So I come from a tech background, so I'm not reluctant to use tech, but it's more towards like I haven't been using tech for like go deep dive into it for like maybe six or seven years.
00:09:03.519 --> 00:09:04.720
So the coding in mind.
00:09:05.200 --> 00:09:05.679
Yeah, yeah.
00:09:05.759 --> 00:09:15.840
So I mean, okay, then so yeah, I I installed everything, use the CRI, and then find out the basics, and then yeah, and then that's where it actually works.
00:09:16.000 --> 00:09:24.000
So to have uh the reports and everything yeah generated and finally understanding like what cloud code actually brings about.
00:09:24.159 --> 00:09:28.720
So I think that was the the whole journey of how I entered up into AI.
00:09:28.799 --> 00:09:56.799
It's it's not a glamorous kind of thing, but it's more towards like out of necessity, and that curiosity actually gave me that opportunity to really know what AI is all about and how we can actually like automate so much of these kind of processes that yeah, it it that if you don't go into it, you literally think that yeah, so AI is just answering a simple question and that's it.
00:09:56.960 --> 00:09:57.600
Yeah.
00:09:59.039 --> 00:10:10.320
We do have a series of you know six videos where we sat down with professionals from like finance to uh wet market and to business owner uh on Claude Code.
00:10:10.480 --> 00:10:18.639
How did Claude Code like massively improve your your outputs uh as compared to like using Chat GPT and Claude?
00:10:18.879 --> 00:10:29.279
And then did you also I think you also dove quite deep into Hermes, which I haven't heard you mention yet, uh into like the open claw type of harnesses.
00:10:29.519 --> 00:10:34.320
Do you mind just touching a bit more on like what did Cloud Code get you and like further?
00:10:34.480 --> 00:10:36.159
Let's go further than that.
00:10:37.279 --> 00:10:40.879
I think uh yeah, so Cloud Code was the start.
00:10:41.120 --> 00:10:42.240
Um yeah.
00:10:42.480 --> 00:10:47.519
I think the Hermes part was uh I definitely have to start with uh OpenClaw.
00:10:47.600 --> 00:10:53.360
So um because OpenClaw is one of the pioneers for agentique uh AI and things like that.
00:10:53.600 --> 00:10:58.399
So definitely that was the first one because the hype was so overwhelming.
00:10:58.639 --> 00:11:05.120
So I needed to know so what does OpenClaw actually give that uh cloud code doesn't?
00:11:05.360 --> 00:11:10.080
So yeah, I have it installed and then uh yeah, to configure it.
00:11:10.159 --> 00:11:15.919
Uh to be honest, it took me more than a day because of all the the configurations and things like that.
00:11:16.080 --> 00:11:31.039
But when when it starts to work, I think uh I finally understand so what agentic programming is, what agent t ai is, because the the moment you have agents working for you uh it's totally different from cloud code.
00:11:31.120 --> 00:11:50.799
Because cloud code, yes, you can set uh uh jobs and things like that, and as especially with cloud co-work and things like that, but to have an agent who has his a mind of his own if you configure it correctly, I think literally it can be uh um let's say an employee that you have.
00:11:51.200 --> 00:12:03.759
So an employee that actually like prompts you, an employee that helps you to do work while you are away, and an employee that actually like um to remind you about things that you may forget.
00:12:04.000 --> 00:12:19.200
So that's where um yeah, OpenClaw gave me that that whole um idea of what agentic AI is, and where that change came about was that when I go into OpenClaw, I don't think it was right from the beginning, it was like in the middle.
00:12:19.519 --> 00:12:26.480
So Hermes was actually launched around that period.
00:12:26.720 --> 00:12:35.039
So when I got into uh open closer, uh I was happy with everything, so I set up with uh all the soul MD and created everything nicely.
00:12:35.279 --> 00:12:39.360
So there was one day that uh okay, there's an upgrade available.
00:12:39.759 --> 00:12:54.000
So I clicked on upgrade, everything died, and it took me like six to eight hours to resolve, even with cloud code on the back, trying to resolve it, uh all the bugs and things like that.
00:12:54.080 --> 00:12:57.919
So that was where I like during the the cloud code debugging, okay.
00:12:58.000 --> 00:13:01.279
Yeah, so Hermes is available, so it's also similar.
00:13:01.360 --> 00:13:02.879
So why don't I try that?
00:13:03.600 --> 00:13:09.200
Okay, so yeah, that's where uh yeah, and my transition to Hermes actually came from there.
00:13:09.440 --> 00:13:18.720
So uh yeah, so I think that that was one of the push, and that was made me that what made me um like look into what Hermes is about.
00:13:18.960 --> 00:13:24.720
So once I used Hermes, basically I ported everything over, so Sol MD and things like that.
00:13:24.879 --> 00:13:37.840
Um for a start, I think there were lesser plug-ins, lesser skills uh available because everything was all open claw, but as the time goes on, uh a lot more of those features actually come into Hermes.
00:13:38.159 --> 00:13:42.000
So I think that's where um I was still keeping like uh tabs on open claw.
00:13:42.080 --> 00:13:43.279
Okay, so what's the difference?
00:13:43.440 --> 00:13:49.919
But I think that comes to a stage whereby they were they were on equivalent, and probably now I think um, yeah.
00:13:50.080 --> 00:14:05.279
To me, actually, Hermes is right at the top level right now because there are certain skills whereby uh like computer use, like browser use, and things like that, it really saves a lot in terms of let's say the the tokens involved and things like that.
00:14:05.440 --> 00:14:07.200
So it's very much optimized.
00:14:07.360 --> 00:14:19.679
And one thing that I like about Hermes is that whenever there's an update available, I will just click on it, so I don't have that fear of yeah, so there's an update.
00:14:19.840 --> 00:14:21.039
So should I update?
00:14:21.120 --> 00:14:28.799
Uh do I have let's say two or three hours available today before I update this because I know that it's gonna break for in terms of open flaw.
00:14:28.879 --> 00:14:29.360
Yeah.
00:14:30.480 --> 00:14:30.799
Yeah.
00:14:30.960 --> 00:14:31.440
Yeah.
00:14:31.679 --> 00:14:31.840
Okay.
00:14:32.639 --> 00:14:35.440
What kind of business were you running before this?
00:14:35.519 --> 00:14:38.240
Like how big was it if you could just give a general scale?
00:14:38.320 --> 00:14:40.399
You don't have to give like exact numbers.
00:14:41.279 --> 00:14:43.759
For omnipotence, yeah, yeah, yeah.
00:14:44.399 --> 00:14:48.559
Um it's uh eight-figure annual revenue business.
00:14:48.799 --> 00:14:56.960
Yeah, so I think in terms of uh scale itself, so we have uh at the peak itself around 40 employees.
00:14:57.200 --> 00:15:01.759
Um on my end, I think there's like seven direct reports.
00:15:02.000 --> 00:15:05.919
So I think that's very much on like yeah, on a day-to-day basis.
00:15:06.080 --> 00:15:08.639
I think that's something that I want to show later on.
00:15:08.799 --> 00:15:13.200
Uh is more to the daily reports that I'm getting from them and things like that.
00:15:13.279 --> 00:15:26.639
So how I actually make use of uh Hermes right now to actually convert whatever that I asked from the my seven employees, and it can actually generate it within uh the morning every morning now.
00:15:27.039 --> 00:15:27.440
Yeah.
00:15:27.759 --> 00:15:28.320
Cool, cool.
00:15:28.559 --> 00:15:31.519
Can you do that now for first press, right?
00:15:31.759 --> 00:15:38.480
Uh yes, so for first press, and I do have another um two possible startups.
00:15:38.639 --> 00:15:51.840
So that's why I have uh Hermes to actually like manage my time because it's more towards like yeah, we I need to manage the time in terms of like um yeah, what day should be dedicated to which areas and things like that.
00:15:51.919 --> 00:15:52.480
Yeah.
00:15:54.559 --> 00:15:59.519
Sorry, one last thing before we get into your stack is how easy was it to switch from open cloud to Hermes?
00:16:00.559 --> 00:16:06.559
Um right now there's actually uh import-export kind of thing, so it's quite straightforward.
00:16:07.120 --> 00:16:09.120
Yeah, so it's a one-click kind of thing.
00:16:09.200 --> 00:16:14.799
Let's say if you want to move from uh open cloud to Hermes, or actually the other way around, it's also available.
00:16:14.879 --> 00:16:15.519
Yeah.
00:16:15.840 --> 00:16:20.320
So they make it quite easy took me uh half a night last week.
00:16:20.639 --> 00:16:20.879
Yeah.
00:16:21.039 --> 00:16:21.200
Yeah.
00:16:21.600 --> 00:16:23.840
Finally took the bullet, yeah.
00:16:24.720 --> 00:16:29.120
Yeah, I think uh too much updates breaking.
00:16:29.279 --> 00:16:30.639
Although I lasted six months.
00:16:31.919 --> 00:16:33.039
Patching every month, right?
00:16:33.120 --> 00:16:35.360
So yeah, until it's it's just too much.
00:16:35.440 --> 00:16:36.799
Yeah, so I decided to change.
00:16:37.120 --> 00:16:38.080
So far, so good.
00:16:38.159 --> 00:16:43.440
I managed to replicate everything, and token use is a bit higher, so I'm trying to optimize more.
00:16:43.759 --> 00:16:44.320
Yeah.
00:16:44.639 --> 00:16:44.960
Okay.
00:16:45.840 --> 00:16:51.120
Are there any maybe if there are any quick comparisons that you guys could call out as well, that would be quite helpful.
00:16:51.200 --> 00:16:53.919
Like I think OpenClaw has more channels you can use it on.
00:16:54.240 --> 00:16:56.000
Hermis a bit less.
00:16:56.159 --> 00:17:02.320
Um, but OpenClaw at least used to break a bit more, Hermes uh seems quite stable.
00:17:02.559 --> 00:17:09.839
And token efficiency-wise, Ben, I think you mentioned you had a better experience with Hermis, and but for OpenClaw not so much.
00:17:09.920 --> 00:17:18.240
Like are there any different are there other considerations you guys were thinking of that you know makes them different?
00:17:18.400 --> 00:17:22.160
As in just to kind of draw some quick distinctions.
00:17:23.039 --> 00:17:31.519
I would say I haven't used OpenClaw enough to understand let's say the the speed of like probably what you're actually experienced.
00:17:31.759 --> 00:17:40.960
So like for a same kind of scenario, so how much time is actually needed from open claw and then compared to Hermes?
00:17:41.519 --> 00:17:48.400
Because most of the time the jobs that uh I run is all on the back end, so instead of me prompting.
00:17:48.480 --> 00:18:06.160
So even for now, I think it is it's probably like uh not so much of a an on an issue that I see in terms of speed, but I think if you talk about uh token usage, um I think it was high at a certain uh certain time, but when I switched over to Luna, yeah.
00:18:06.240 --> 00:18:17.359
So I think Luna Luna was one of the uh biggest um changes that OpenAI came up with, so with that lower cost and everything, even now, let's say
for the free chat GPT, right?
00:18:17.440 --> 00:18:19.680
They're all using Luna and it's unlimited.
00:18:19.920 --> 00:18:23.839
So I think that's that's uh that's a big change.
00:18:24.079 --> 00:18:25.599
And you'll be surprised.
00:18:25.680 --> 00:18:28.880
Like so, for me, I've been using Luna for the past one week.
00:18:29.039 --> 00:18:38.319
Uh the quality that it actually produces for agent tasks, I think is yeah, it's really comparable to any kind of other LLMs.
00:18:38.400 --> 00:18:38.720
Yeah.
00:18:39.039 --> 00:18:47.680
Previously I was using uh DeepSeek for the for the uh lesser important task, but I think Luna is much much better now.
00:18:48.000 --> 00:18:53.839
So Luna is the fastest and least smart version of GPT 5.6.
00:18:54.160 --> 00:18:55.039
Is that accurate?
00:18:55.599 --> 00:18:57.119
Yes, correct.
00:18:57.359 --> 00:18:57.839
Yeah, yeah.
00:18:58.559 --> 00:19:04.480
Uh the the hack is that you don't use the the mid version, you just use the max of the Luna.
00:19:06.079 --> 00:19:06.319
Okay.
00:19:06.559 --> 00:19:08.720
So effort is you, yeah, correct.
00:19:08.880 --> 00:19:09.599
You use MIT.
00:19:10.319 --> 00:19:12.000
Maybe maybe Ben back to you.
00:19:12.079 --> 00:19:17.440
Um what's the main AI stack you currently use, you know, before showing us your screen?
00:19:18.480 --> 00:19:20.400
Um cloud definitely.
00:19:20.480 --> 00:19:24.000
So I'm how much time do you spend in each of them if you could just share as well?
00:19:24.559 --> 00:19:26.400
My stack for AI, I think quite simple.
00:19:26.480 --> 00:19:29.440
I use cloud code and also the cloud AI version of it.
00:19:29.599 --> 00:19:32.799
Um I don't use the API at all because it's too expensive.
00:19:33.039 --> 00:19:39.680
So cloud code, I think, yeah, basically everything from coding to research to everything.
00:19:40.000 --> 00:19:44.960
I guess most of the work that uh Ben you do is non-technical, would you say?
00:19:45.119 --> 00:19:49.039
Because you're mostly running a business with AI or are you coding quite a lot as well?
00:19:49.599 --> 00:19:53.039
Um I do it probably on a 6040 kind of thing.
00:19:53.279 --> 00:19:53.599
Okay.
00:19:53.839 --> 00:19:57.039
So yeah, mostly it's non-coding.
00:19:57.359 --> 00:20:04.720
Maybe Ben, if you could share a bit about your use cases or or maybe any case studies uh that you wanted to share.
00:20:05.759 --> 00:20:09.440
So um we talk about my AI stack.
00:20:09.519 --> 00:20:20.000
So my AI stack consists of let's say cloud code, cloud AI, and very much Hermes, and sometimes also like codecs or ChatGPT.
00:20:20.319 --> 00:20:37.200
So when you're running so many things at the same time, um there comes to a point whereby uh let's say if you do something on uh Hermes, uh you ask Hermes something or you ask Hermes to remember something, everything will be stuck on Hermes.
00:20:38.480 --> 00:20:49.200
So when you use cloud code to try to okay, expand on that particular um subject, basically he has no knowledge of what Hermes actually came up with.
00:20:49.759 --> 00:21:05.279
So today, what I'm trying to show you here is more towards like so how do you make like the different kind of tools that you use um be able to talk to each other in in a way whereby yeah they have a shared kind of context.
00:21:05.519 --> 00:21:20.559
So the moment you um remember something for let's say uh in Hermes, basically, when you look at cloud code itself, you'll be able to extract whatever discussion you have and then yeah work from there again.
00:21:21.440 --> 00:21:25.440
Yeah, so this is something that uh I have so far.
00:21:25.599 --> 00:21:32.079
So um what I have here is more towards the the agent setup.
00:21:32.480 --> 00:21:33.680
So this is my Hermes.
00:21:34.160 --> 00:21:39.359
So Hermes is split up into two different agents, so I keep it very simple.
00:21:39.519 --> 00:21:59.839
So one agent is called Holly, so very much on the household schedules, so because I have two kids, so uh to tell me like okay, so what kind of extracurricular activities they have, um do they have school today, uh if I have an environment to pick them up, so does it clash with my other meetings?
00:22:00.240 --> 00:22:08.079
Things like that, and it also helps me to find out like uh okay, recovering payments, remind me to make payments and things like that.
00:22:08.319 --> 00:22:11.519
So that's my household agent.
00:22:11.759 --> 00:22:17.200
Then the next agent, I think, is the key thing, and I think I'll be talking a lot more on this agent.
00:22:17.359 --> 00:22:19.200
So uh is Mimi.
00:22:19.599 --> 00:22:21.200
So that's my chief of staff.
00:22:21.440 --> 00:22:26.480
So basically, this one is like my so-called discipline master.
00:22:26.799 --> 00:22:36.480
So she's the one that actually looks at all my external kind of businesses, uh, making sure that um my targets are being met.
00:22:36.720 --> 00:22:43.519
So things like uh hey, I need to make sure that I have this business up and running by September.
00:22:43.759 --> 00:22:47.119
So she'll be the one that keeps reminding me on a daily basis, okay.
00:22:47.200 --> 00:22:51.039
Hey, your deadline is like um 30 days away.
00:22:51.279 --> 00:22:53.200
Have you done this, this, this, this, this?
00:22:53.519 --> 00:22:55.680
So that is the role of Mimi.
00:22:56.720 --> 00:23:01.279
Okay, so and um this is only the the tip of the iceberg.
00:23:01.519 --> 00:23:08.079
So having Hermes here, um, I do have actually like um an always on iMac.
00:23:08.240 --> 00:23:19.599
So Hermes is actually on the Always on iMac, and what I have on my MacBook Air is that yeah, it is a portable one, so that's where all my cloud uh cloud code everything lives.
00:23:19.839 --> 00:23:32.640
So that's the kind of infrastructure that I have, and it's challenging, so I'm not trying to showcase in this entire session here, but uh yeah, this is a high-level one that that actually I can show.
00:23:32.880 --> 00:23:52.640
Okay, so how all this works is that um on a daily basis, what I have uh Mimi and Holly do is that so it will actually prompt me on a daily basis in terms of like um using Telegram, so at different times of the day.
00:23:52.799 --> 00:23:54.160
So, firstly is my 8 a.m.
00:23:54.400 --> 00:23:54.880
briefing.
00:23:55.039 --> 00:23:58.079
So, what are the five important things that needs to be done?
00:23:58.160 --> 00:24:11.039
What are the things that uh you did not do, and everything, and then there are also like a lot more of um uh other reports as well.
00:24:11.200 --> 00:24:25.519
So basically, I'm still like tracking Omnidask kind of performance, so it'll be giving me like daily reports and then weekly reports, um, all coming off from the data being extracted on the into my context layer.
00:24:25.680 --> 00:24:32.160
So this is where um the telegram works, and then finally, when how does it all work?
00:24:32.319 --> 00:24:33.759
It comes down to this.
00:24:34.079 --> 00:24:36.400
So my context layer is here.
00:24:37.200 --> 00:24:42.720
So this is a very broken down or rather simplified version.
00:24:42.880 --> 00:25:00.319
So there are four different kinds of um files that I have that whatever sessions that I create, let's say on uh Hermes or on Cloud Code, it will read all these four files, and that's where all the kind of context actually comes into.
00:25:00.640 --> 00:25:08.319
And how does this all four files uh get all those information is where the data CLI comes in?
00:25:10.559 --> 00:25:12.000
Data CLI.
00:25:12.400 --> 00:25:17.200
Uh data or um any kind of APIs or CLIs, everything.
00:25:17.359 --> 00:25:29.119
So I try to use like CLIs as much as possible so that uh yeah the data can be stored inside the database uh in my in my uh in my iMac itself.
00:25:29.200 --> 00:25:29.440
Yeah.
00:25:29.759 --> 00:25:40.240
So by that you mean uh for your agent to kind of use the CLI tools to get the necessary info they need from tools like Shopify, Meta.
00:25:40.640 --> 00:25:41.039
Yep.
00:25:41.279 --> 00:25:41.440
Okay.
00:25:41.759 --> 00:25:42.000
Correct.
00:25:42.160 --> 00:25:42.559
Yep.
00:25:43.440 --> 00:25:47.039
So you run a daily job that pulls the data out?
00:25:47.359 --> 00:25:47.759
Uh yes.
00:25:48.079 --> 00:25:51.119
It says it in the C correct yeah.
00:25:51.599 --> 00:26:01.039
So I have a uh very simple simple kind of database in the iMac itself to store all the Shopify data, clue your data on a daily basis.
00:26:01.119 --> 00:26:03.599
So every day will have a line item itself.
00:26:03.920 --> 00:26:04.559
Yeah.
00:26:07.119 --> 00:26:11.440
Sorry, so what differentiates the iMac from the MacBook Air again?
00:26:12.799 --> 00:26:15.920
Okay, so uh the iMac is always on.
00:26:16.240 --> 00:26:16.880
Yep, okay.
00:26:17.279 --> 00:26:18.319
So that's for your Hermes.
00:26:18.799 --> 00:26:26.960
Yes, because Hermes, what it does is that uh it needs to keep running uh every single job there, so without breaking.
00:26:28.240 --> 00:26:29.839
And then the MacBook Air?
00:26:30.240 --> 00:26:38.240
So the MacBook Air is my workhorse, so that's where I usually do all my um cloud code stuff and everything here.
00:26:38.720 --> 00:26:45.920
Because then uh yeah, I I still travel a lot with it, so like I can do it on the outside, I can do it anywhere else.
00:26:46.079 --> 00:27:00.720
Um that's where everything, and then let's say if I need to access data, that's where I use Hermes on my MacBook Air to connect back into my Always On and get all the whatever context that I need.
00:27:03.200 --> 00:27:17.359
So maybe um for those that are just listening and can't see what's on the screen, um on the left of uh Ben's screen that shows the data layers, so that's like Shopify, uh Meta Ads, uh his notes.
00:27:17.519 --> 00:27:18.480
I guess Obsidian.
00:27:18.559 --> 00:27:19.519
Is that Obsidian?
00:27:19.839 --> 00:27:20.799
Yes, correct.
00:27:21.119 --> 00:27:31.440
Okay, okay, and then uh it feeds into a context layer with the state.md, with the priorities.md, with the decisions.md and the voice.md.
00:27:31.759 --> 00:27:37.519
Uh and then that feeds into the two agents, Holly and Mimi, who are not real humans.
00:27:39.119 --> 00:27:47.680
Um and they basically have um different uh cron jobs, I guess, like an 8 a.m.
00:27:47.839 --> 00:27:53.759
briefing, uh there's a telegram connection as well and a weekly report.
00:27:54.079 --> 00:28:24.000
Um and these all run on the iMac, which is an always-on uh Hermes, basically always on iMac that runs Hermes, and then um Ben you also use your MacBook Air to basically take things out of this iMac uh when you need to continue working on maybe a piece of work that the Hermes agent does not have context of.
00:28:24.079 --> 00:28:26.400
Is that a good summary of what I'm seeing?
00:28:26.640 --> 00:28:27.359
Yes, correct.
00:28:27.519 --> 00:28:27.920
Yep.
00:28:28.400 --> 00:28:29.440
Okay, okay.
00:28:29.680 --> 00:28:39.119
And for the context layer, was that recommended by Hermes or was that something that you kind of uh tweaked as you continue to experiment?
00:28:39.759 --> 00:28:43.279
So the context layer is really tweaked and experiment.
00:28:43.599 --> 00:28:56.160
So this is a very simplified version, um, which in uh in my opinion, let's say anybody who hasn't um started should use this um simplified version.
00:28:56.480 --> 00:29:01.680
Because I did use a lot of the oh the second brain setups and things like that.
00:29:01.920 --> 00:29:08.559
So uh even from the the YC CEO uh when he came up with that one.
00:29:08.720 --> 00:29:10.880
Um Gary Tonner, yeah.
00:29:11.039 --> 00:29:11.599
Yes, Gary.
00:29:11.839 --> 00:29:19.039
So when when you use those like so-called templates, it will be very useful when they showcase it.
00:29:19.359 --> 00:29:29.200
But when it comes down to us, I think everyone has a different kind of layer, everyone has different
businesses, everyone has different kinds of processes.
00:29:29.680 --> 00:29:40.079
So you would want to have your own kind of context layer rather than building off something that is so stacked up.
00:29:40.559 --> 00:29:54.160
So when when you you have a context layer for coders, when you have a context layer for business owners or or maybe even housewife, basically they have very different kind of context that they want to keep.
00:29:54.400 --> 00:30:10.400
So if you're gonna uh use a second brain, let's say uh downloaded or uh using Gary's one, I think it's really too much to start, and you get like um overwhelmed by it and to the stage whereby you just stopped using it.
00:30:10.640 --> 00:30:16.640
So that that's something that a lot of people have have bumped into, and a lot of my friends have actually got into this trouble.
00:30:16.720 --> 00:30:19.839
It's like, hey, I want to use second brain, why is second brain so good?
00:30:19.920 --> 00:30:26.640
Uh I download that, but it doesn't work, it keeps break, it keeps breaking, it doesn't like give me like what I want.
00:30:26.960 --> 00:30:37.279
So it's more towards like instead of like trying to come up with like so uh or follow such a stringent convoluted kind of context layer.
00:30:37.440 --> 00:30:39.440
So it's always better to start it yourself.
00:30:39.519 --> 00:30:43.680
So I hope that this uh particular one which is broken down.
00:30:43.839 --> 00:30:52.799
Um I've seen this on uh on X quite a couple of times, so which is a very simplified version of something uh for second brain.
00:30:53.039 --> 00:30:59.599
So these are the key things whereby it will help you to create your own context loop.
00:30:59.839 --> 00:31:12.400
So once you understand this, then that's where you actually expand further, and to actually build on this rather than uh to cut down on a uh convoluted one, I think it's much much easier.
00:31:12.559 --> 00:31:16.960
So that's the that's the thing about um these four different files here.
00:31:18.160 --> 00:31:19.759
So so I'm actually quite curious.
00:31:20.160 --> 00:31:25.920
Does this list just keep appending new things or do you clean it up, you know, like priorities?
00:31:26.240 --> 00:31:29.279
Do you store the old ones or do you clean it up and keep it somewhere else?
00:31:29.839 --> 00:31:30.960
And decisions also, right?
00:31:31.119 --> 00:31:33.920
I suppose that you just add on more over time.
00:31:34.000 --> 00:31:34.160
Yeah.
00:31:34.400 --> 00:31:36.559
Do we do we want to jump into this one first?
00:31:36.960 --> 00:31:39.440
I can I can I can let you see here.
00:31:39.680 --> 00:31:43.680
So uh so this is the entire pack.
00:31:43.839 --> 00:31:46.960
So we have the sorry, what is this pack?
00:31:47.759 --> 00:31:52.079
Okay, so this is the operator context pack, which is the one that I'm gonna share later.
00:31:52.480 --> 00:31:54.640
So this is whatever you see here.
00:31:54.720 --> 00:31:55.440
So these four files.
00:31:56.000 --> 00:31:57.279
Okay, okay, okay, okay.
00:31:57.680 --> 00:31:59.039
So I have it over here.
00:31:59.119 --> 00:32:00.960
So the four files are all here.
00:32:02.640 --> 00:32:09.920
So to answer Xiao Hong's question first, I do have a skill which is called the weekly review skill.
00:32:11.839 --> 00:32:13.200
So this is what it does.
00:32:13.279 --> 00:32:24.640
So it will actually read all the files, and then looking at your work log, so what you actually did for the entire week, that's where it will actually reconcile everything.
00:32:24.799 --> 00:32:28.240
So, did you actually work according to your priorities?
00:32:28.400 --> 00:32:38.640
Uh is there any updates to the priorities, and it will actually be asking you so should the priorities be changed once you execute this particular skill itself?
00:32:39.039 --> 00:32:45.039
So, yes, it will actually keep updating, and yeah, that's why I call it a context loop.
00:32:45.200 --> 00:32:52.559
So, if you have a um a context layer that doesn't change, then that's gonna be a problem.
00:32:52.640 --> 00:33:01.599
So, which is what a lot of uh other uh like second brain um kind of like skill sets, there is is actually doing.
00:33:01.759 --> 00:33:15.440
So they have everything perfectly done up, but the fact is that they wanted people to update their own priorities list, they needed people to update what they did in the MD files before it generates everything nicely.
00:33:15.680 --> 00:33:28.799
So if users don't actually do that manually or they don't know how to automate that on their own, that's where it actually gets lost because they say, Hey, I have this here, I have my task done, but it's not updating.
00:33:29.039 --> 00:33:35.599
Yeah, so this is the the skill that actually like makes it into a loop whereby it updates every single one over here.
00:33:39.359 --> 00:33:39.759
Interesting.
00:33:40.319 --> 00:33:45.119
So you actually score yourself on how well you worked, yes, correct.
00:33:45.279 --> 00:33:46.079
Yeah, yeah.
00:33:46.400 --> 00:33:48.480
And what do you do with the information?
00:33:48.880 --> 00:33:49.759
What better?
00:33:50.559 --> 00:33:57.039
Uh it so there's um the part whereby it goes back to Hermes itself.
00:33:57.519 --> 00:34:14.000
So when we when we have all these like uh priorities decisions made and then we do a a weekly review, so I do have uh the the weekly briefing um from Hermes to say that okay, hey, um this is the weekly review, what have you done so far?
00:34:14.239 --> 00:34:19.199
Is this based on what you actually agreed on in terms of last week's priorities?
00:34:19.440 --> 00:34:24.559
So to me that is very important because when Hermes prompts me at 8 a.m.
00:34:24.800 --> 00:34:33.119
every day, so uh Mimi, uh my chief of staff, is very straightforward.
00:34:33.679 --> 00:34:49.280
So they would he will say that okay, it is a bad week for you, you didn't hit your target, your pipeline is zero, um you don't have
any kind of um coffee meetings with anybody next week, you need to get one now, and things like that.
00:34:49.360 --> 00:34:54.159
So this is how uh it actually pushes me to okay, yeah, okay.
00:34:54.400 --> 00:34:58.079
So I set out a goal last week, I didn't meet.
00:34:58.320 --> 00:35:02.480
Yeah, these are the things that need to be done, that I need to have done right now.
00:35:02.639 --> 00:35:03.280
Yeah.
00:35:04.719 --> 00:35:05.119
Interesting.
00:35:05.840 --> 00:35:09.360
Technically, if you use um password managers, you can.
00:35:10.480 --> 00:35:10.719
Yeah.
00:35:11.280 --> 00:35:11.599
Yeah.
00:35:11.760 --> 00:35:15.679
So I I actually use one password and then I set up a service account.
00:35:16.719 --> 00:35:24.000
And it only has a VOD that um the agent has access to, so only specific uh logins for it.
00:35:24.400 --> 00:35:26.079
So it doesn't read like the rest of my stuff.
00:35:26.159 --> 00:35:26.559
Yeah.
00:35:27.440 --> 00:35:28.400
So you actually can do it.
00:35:28.480 --> 00:35:28.639
Yeah.
00:35:28.960 --> 00:35:29.199
Yeah.
00:35:29.280 --> 00:35:34.400
And and maybe even beyond the passwords and security is also just like context, right?
00:35:34.559 --> 00:35:39.679
Because the like I I work quite a lot in cloud code as well.
00:35:39.840 --> 00:35:46.480
Um and then of course the ad hoc stuff goes to my my Hermes slash open claw.
00:35:46.719 --> 00:35:51.679
So Ben, do you do you kind of update your context when you're working on cloud then?
00:35:51.760 --> 00:35:51.920
Yeah.
00:35:52.079 --> 00:35:52.320
Yes.
00:35:52.559 --> 00:35:53.920
Based on what every is saying, yeah.
00:35:54.159 --> 00:35:54.480
Yeah.
00:35:54.639 --> 00:35:59.679
So how cloud does uh its update is through the hooks itself.
00:36:00.079 --> 00:36:06.719
So uh my one is a bit over overboard, uh so you can actually do it on let's say um manually.
00:36:06.960 --> 00:36:18.719
So let's uh if you have that whole conversation that and you want to log it into your uh worklock, then basic basically you just create a skill called uh log work, then it will just go into your notes and box.
00:36:19.920 --> 00:36:24.960
Okay, but for me is that uh every turn I have my hook that actually does it.
00:36:25.119 --> 00:36:30.480
So every time it's like uh whatever question, whether correct or not, right, everything will be locked into it.
00:36:30.719 --> 00:36:36.400
So my basically my notes and that that whole work log is actually quite quite piled up.
00:36:36.960 --> 00:36:38.079
Wow, it's quite heavy.
00:36:38.239 --> 00:36:38.320
Yeah.
00:36:38.480 --> 00:36:39.920
And where do you saw that?
00:36:40.239 --> 00:36:40.960
Markdown force.
00:36:41.199 --> 00:36:43.360
It's all in it's all in obsidian.
00:36:43.599 --> 00:36:44.880
Oh obsidian, okay, okay.
00:36:45.199 --> 00:36:46.800
So it's all marked down for the same thing.
00:36:46.880 --> 00:36:47.039
Yeah.
00:36:47.679 --> 00:36:48.719
Yeah, yeah.
00:36:49.199 --> 00:36:49.920
I
don't know.
00:36:50.000 --> 00:36:52.480
I I I think it's super interesting and very cool.
00:36:52.639 --> 00:36:57.519
Um does your obs do you ever feel like your obsidian is exploding?
00:36:58.880 --> 00:37:00.960
Uh I feel that it's not filled up enough.
00:37:01.119 --> 00:37:03.360
Because um these workloads, right?
00:37:03.440 --> 00:37:08.800
I I don't even like surface them out on obsidian.
00:37:08.880 --> 00:37:11.119
So I hide I hide the entire folder.
00:37:11.440 --> 00:37:11.679
Okay.
00:37:12.079 --> 00:37:18.000
Yeah, so it my obsidian vault is only all the critical uh notes and information that is required.
00:37:18.079 --> 00:37:19.280
Uh that is readable.
00:37:19.760 --> 00:37:24.559
Anything that's non-readable, I just hide them because it it would if not you just clutter your whole obsidian.
00:37:24.719 --> 00:37:25.039
Yeah.
00:37:26.559 --> 00:37:33.280
Yeah, because when I think obsidian, I think of second brain, and when I think second brain, I think my whole life.
00:37:33.920 --> 00:37:36.800
So so yeah, my obsidian is very cluttered.
00:37:37.039 --> 00:37:42.960
Um but it is a useful place to like pluck out information from like 10 years ago about myself.
00:37:43.199 --> 00:37:48.960
Um but yeah, not not uh not a big security advocate here, but yeah.
00:37:49.199 --> 00:37:52.880
Uh so obsidian is definitely uh for an individual, right?
00:37:52.960 --> 00:38:00.559
So the way this whole setup is meant to be individual when it comes to a business, how do you intend to then expand this out?
00:38:01.760 --> 00:38:09.679
I think the the key thing that that was what I was uh thinking about, and uh I'll be upfront, I'm not a Notion fan.
00:38:09.920 --> 00:38:21.199
So I know everything can be put into Notion, but to me, oh Notion is too much of a bells and whistles, and it takes a lot of effort to actually go into it.
00:38:21.440 --> 00:38:24.079
So that's something that I always wanted to avoid.
00:38:24.320 --> 00:38:38.639
So right now I think that's that's still the question mark that uh I'm not sure if let's say uh one of the businesses has a bigger team and things like that, but I think recently there's something that came up.
00:38:38.960 --> 00:38:47.519
So um the ex Twitter boss, uh Jack, so he came up with something called bus.
00:38:48.400 --> 00:38:49.679
Buzz.xyz.
00:38:50.239 --> 00:38:56.239
So it's literally a slack kind of um yeah, replica.
00:38:56.480 --> 00:39:01.840
So, but what it does is that it includes agents and also humans within it.
00:39:02.480 --> 00:39:05.920
It's uh very now, we are testing it.
00:39:06.320 --> 00:39:08.719
That is one of the first projects that I did with Cloud Code.
00:39:08.880 --> 00:39:09.360
Alright, okay.
00:39:09.519 --> 00:39:10.000
Hi there.
00:39:10.159 --> 00:39:12.159
I'm so glad you found this podcast.
00:39:12.320 --> 00:39:24.159
If you'd like to support us to help others flourish and have fun with AI, please consider subscribing, giving us a like,
review, or comment on wherever you listen to your podcast, as that helps us keep the show running.
00:39:24.480 --> 00:39:28.239
I don't know if you have seen um how Shopify POS works.
00:39:28.400 --> 00:39:33.440
Yeah, so Shopify POS is a very simplified version of uh POS system.
00:39:33.519 --> 00:39:42.079
So whenever a customer wants to buy something, you can go in and basically select the SQU that they want to buy and add into their card.
00:39:42.159 --> 00:39:44.880
So it's like uh item by item kind of process.
00:39:45.039 --> 00:39:50.800
So in OmniDesk itself, as you know, uh if you go to our website, um it's all about configuration.
00:39:50.960 --> 00:39:53.280
So we can uh have a desk, right?
00:39:53.440 --> 00:39:57.920
So when you buy a desk, there are several different kinds of SQUs available.
00:39:58.079 --> 00:40:05.360
So we have the tabletop, we have the frames, we have the uh cable management bar, that's the minimum, and then you add on different items.
00:40:05.599 --> 00:40:33.599
So the problem we had with Shopify POS for the past six years or so is that whenever I have a part-time staff like doing uh trade shows, road shows, and things like that, uh coming in to do sales, I cannot have them use the POS because that error, um, how do you say the rate of error is gonna be very high because there's we have like 500 over SKUs, and I can't have them like selecting one by one based on what customer chose.
00:40:33.760 --> 00:40:40.880
So let's say if Yahoo wants to buy uh uh ominous ascent today, so basically the customer is just gonna tell you, so I want to buy an ominis ascent.
00:40:41.199 --> 00:40:48.159
I'm not gonna tell you what's the SKU, so unless they go and match it up uh with our documentation, they can't do it in the POS.
00:40:48.239 --> 00:40:49.840
So this is what I created.
00:40:50.000 --> 00:41:03.199
So it's a POS extension which actually shows uh allows any uh like any staff like to be able to build a desk similar to how the flow comes out inside our website right now.
00:41:04.000 --> 00:41:12.000
So you build a phone app yes, so it's it's built on the Shopify POS itself, so it's a uh so-called extension app.
00:41:12.639 --> 00:41:12.960
Okay.
00:41:13.920 --> 00:41:17.599
And you call the API uh yes, correct on Shopify, yeah.
00:41:17.760 --> 00:41:25.360
Yeah, so this is the when you see here the this build a desk, so this is the so-called uh app extension that I created.
00:41:26.480 --> 00:41:35.440
So when anybody goes in, so when they click on build a desk itself, so they can actually select the model.
00:41:35.840 --> 00:41:43.199
So whether is it uh the ascend thing uh the ascend zero, select the frame, and then select the tabletop.
00:41:43.519 --> 00:41:55.760
So all these are based on let's say the criteria, and then it appears uh based on let's say the let's say if you have a black frame, you cannot have a different color kind of cable management bar.
00:41:56.000 --> 00:42:01.039
So all
those criteria are all yeah set up in the back end using this particular app.
00:42:01.599 --> 00:42:05.280
Yeah, so by by doing this, right?
00:42:05.360 --> 00:42:12.559
So we actually save like yeah, 50 to 70 percent of the transaction time at Broad Shows.
00:42:12.960 --> 00:42:28.960
So previously, what we have was uh when a customer wants to purchase, like say a thousand five hundred dollars desk, they have to see our sales staff okay, mark off uh entire like literally A4 sheet paper, the SQLs and everything.
00:42:29.199 --> 00:42:33.280
They have to go over with that sheet of paper to our counter.
00:42:33.840 --> 00:42:41.599
The counter person will be keying in all this into Shopify backend and then process the payment.
00:42:42.239 --> 00:42:45.840
So that was the process for almost six years.
00:42:46.159 --> 00:42:57.360
But with this in place, basically we have uh by the time our sales staff clicked through everything with the customer, the customer goes over there just to key in the credit card details, that's it.
00:42:57.519 --> 00:42:57.920
Wow.
00:42:58.079 --> 00:43:05.440
So we cut it from let's say a 15-minute kind of process to probably like a two-minute kind of process.
00:43:05.920 --> 00:43:06.480
Yeah.
00:43:07.199 --> 00:43:11.360
So that's the that's one of the biggest things that we implemented.
00:43:11.440 --> 00:43:12.480
Yeah, yeah.
00:43:13.360 --> 00:43:14.000
Wow.
00:43:14.320 --> 00:43:20.719
I'm guessing it the difficult part was not just building this, you know, button over here, build a desk.
00:43:20.800 --> 00:43:27.760
And the hard part was not probably not you know building the fields to click which table you want, right?
00:43:27.920 --> 00:43:46.000
Like I'm guessing there was probably a lot that went into how you um wired everything up at the back end, and then not just that, but like how you um remove the need for that piece of paper, and then I guess training your staff to use this should be quite easy.
00:43:46.079 --> 00:43:51.440
But like how did you how do you remove the paperwork and and all that?
00:43:51.679 --> 00:43:56.960
Um the the difficult things that happens behind this very simple looking thing.
00:43:58.800 --> 00:44:16.159
Um I think The the hard part I I think for anybody out here is more to yeah, so what does it mean that um if I have a white let's say a white frame I cannot have a white cable management bar uh or I cannot have a black cable management bar.
00:44:16.320 --> 00:44:20.639
So all these like um uh criteria and things like that.
00:44:20.800 --> 00:44:34.079
So the fact is that to me I think that's one of the easiest things uh for this entire app because I've been in this uh for so long, so all these kind of criteria and things like that is all at the back of my head.
00:44:34.400 --> 00:44:46.719
So the hack that I came up with uh and I want to share with uh many people is that so there are a lot of processes within scamp within companies that don't have SOP.
00:44:46.960 --> 00:44:49.920
So I wanted to have this long time ago.
00:44:50.159 --> 00:44:53.199
So let's say uh what can people buy?
00:44:53.360 --> 00:44:58.480
Uh what are the things that this customer cannot have when they choose this particular table and things like that.
00:44:58.559 --> 00:45:02.079
So we always wanted to have that kind of documentation, but we don't have it.
00:45:02.320 --> 00:45:09.840
So when I created this app, I definitely need that, and how I created it was via Whisper.
00:45:12.400 --> 00:45:16.800
So what I did was more to yeah, rambling off.
00:45:17.199 --> 00:45:20.400
You mean you just you just rented, yeah.
00:45:20.960 --> 00:45:26.960
Yes, so ascend is uh so in a in a very simple way, you say.
00:45:27.119 --> 00:45:33.360
So ascend is the top um top tier product, so we have like uh white frame and then we have a black frame.
00:45:33.519 --> 00:45:37.440
So when a customer chooses this particular thing, okay, you cannot have this.
00:45:37.599 --> 00:45:42.800
Uh, if a customer chooses, like say the wow word, so all the wow word versions of the desk will come out.
00:45:42.880 --> 00:45:45.039
So, what consists of a wow word version?
00:45:45.119 --> 00:45:56.800
Uh, these are the tabletop that wow wood, and this is the category that they are in, and basically by rambling this for let's say half an hour or so, uh, I have the full documentation there.
00:45:56.960 --> 00:46:03.599
So that's where I okay, so edit it a bit and yeah, show this entire documentation.
00:46:03.760 --> 00:46:09.599
Okay, so this is the back end that I need to wire to my uh to my this build a desk app.
00:46:10.000 --> 00:46:16.719
So after that, then yeah, please build it and using the latest Shopify um API.
00:46:16.960 --> 00:46:20.800
So I think Shopify has uh because it's it keeps refreshes its API.
00:46:20.880 --> 00:46:31.280
So I think that was the another part that I had to jump through is more to you have to use the Shopify admin uh MCP to make sure that you have the latest kind of APIs.
00:46:31.440 --> 00:46:34.880
If not, you yeah, you'll you bump the door a lot of issues, yeah.
00:46:35.840 --> 00:46:41.599
So yeah, that that was the whole process for for this particular app.
00:46:42.000 --> 00:46:47.840
And you'll be surprised, I think it was just a two-three-day kind of project.
00:46:48.079 --> 00:46:48.639
Yeah, yeah.
00:46:48.880 --> 00:47:04.800
I think we we try I I wanted to do it just the week before uh one of the shows, and yeah, we we did it in within like two days, and then we tested for I told my team to test it for the next three to four days so to make sure that everything is working.
00:47:04.880 --> 00:47:05.519
Yeah.
00:47:08.400 --> 00:47:19.039
And for this, uh I guess you used to have to train your staff on you know what are the different configurations, and then they would kind of pass down that knowledge.
00:47:19.280 --> 00:47:22.480
And and that's where they type it in on the back end.
00:47:22.719 --> 00:47:23.199
Yep.
00:47:23.440 --> 00:47:26.000
Yeah, what what what did it take to remove the piece of paper?
00:47:26.159 --> 00:47:28.079
Was that was that piece of paper important?
00:47:28.800 --> 00:47:35.360
So okay, so why that piece of paper is that so when you go to our roadshows and things like that, right?
00:47:35.760 --> 00:47:41.840
So because our whole business setup is very much on online based.
00:47:42.320 --> 00:47:46.719
So we are catering to cons consumers who are buying online.
00:47:47.280 --> 00:47:53.119
So the best kind of configuration or the best kind of buying process is always through our website.
00:47:53.440 --> 00:48:03.199
So there was a time whereby okay, people were uh at the roadshows, we were locked them in using our computers there to get them to purchase online.
00:48:04.320 --> 00:48:17.199
So that was one of the processes, but uh online has its own online uh issues because customers do not want to key in things uh on a computer and things like that, so we have a lot of issues, so yeah.
00:48:17.280 --> 00:48:24.079
So we did it on the back end so to to make them more comfortable, so that's where the paper comes in.
00:48:24.239 --> 00:48:44.800
The paper comes in because okay, so uh sales staff will record everything down, and then to make like uh customer feel more secure is that they only need to key in whatever uh uh credit card details at a dedicated terminal where our stuff is running, so that's where the the paper comes in.
00:48:44.960 --> 00:48:51.760
So in the end, we have to do it like the the very kind of very manual way in order to appease our customers.
00:48:52.159 --> 00:48:52.559
Yeah.
00:48:53.519 --> 00:48:56.559
It's is to give the bespoke experience.
00:48:57.360 --> 00:49:00.159
Yeah, I mean you can say that, yeah.
00:49:12.559 --> 00:49:19.039
Ben, one last thing I I was wondering was on your dashboards that you are updating every day, right?
00:49:19.519 --> 00:49:22.079
Um how how do you think about building those dashboards?
00:49:22.159 --> 00:49:38.639
And uh I I know maybe you can't share it because there's a lot of company data, but like um like how do you think about choosing the right things to show and how do you wire everything up with the necessary you know numbers that you need to find?
00:49:38.960 --> 00:49:43.760
I can show you something that is probably not that great.
00:49:44.159 --> 00:49:44.480
Okay.
00:49:44.880 --> 00:49:45.280
Yep.
00:49:45.599 --> 00:49:50.880
So so um this is literally like what's going on in my whole infrastructure now.
00:49:51.440 --> 00:49:52.400
But it helps you, right?
00:49:52.639 --> 00:49:54.239
Yeah, it it it does help.
00:49:54.400 --> 00:50:07.599
So the fact is that right, so I'm talking a lot on uh yeah, to have like uh a good context to have everything um properly uh created, and you talk about the dashboards.
00:50:07.679 --> 00:50:19.039
So dashboards I feel that yeah, they are very important, but at the same time, if you look at this uh particular mission control, so this mission control was actually created two months back.
00:50:19.280 --> 00:50:23.840
Okay, it it took me like probably uh yeah, half an hour or so and things like that.
00:50:24.000 --> 00:50:32.320
Um, but the fact is that I haven't been looking at this mission control until I was preparing for this particular podcast.
00:50:32.639 --> 00:50:44.559
So when when I start to look at it, it uh when uh look at it like maybe like a couple of days back, it's like it shows me a lot of issues and it wasn't updated and things like that.
00:50:44.719 --> 00:50:58.239
So I think it it goes boils down to a part whereby yeah, you can have a lot of dashboards, you can have a lot of uh uh uh uh mission controls to to give you that that full vision kind of thing.
00:50:58.559 --> 00:51:19.920
But I think a lot of times we fall into that rabbit hole whereby building infrastructure uh more than doing what we're supposed to do, uh that's the the the bad part about let's say um yeah to to to really say that hey uh yeah I want a very good back end and things like that.
00:51:20.079 --> 00:51:52.320
So the amount of time that you spend and then you will fall into this trap whereby hey I built something and it it actually looks good and it works, but that's what a lot of people right now are uh are falling into is that sometimes it's not about um yeah getting the perfect infrastructure and it's more towards like doing what actually your goals is supposed to be.
00:51:52.480 --> 00:51:53.840
That that is the key.
00:51:54.000 --> 00:51:58.239
So because AI makes it so easy for build things, it's like we we fall into the trap.
00:51:58.320 --> 00:51:59.199
Yeah, we're building things.
00:51:59.280 --> 00:52:04.639
I built like uh to be honest, I think I built like 20 over apps as of to date.
00:52:04.800 --> 00:52:08.480
But the fact is that this 20 over app is that so what does it give me?
00:52:08.639 --> 00:52:16.960
Okay, yeah, maybe some some fun for my kids and things like that, or something that oh wow, I can show someone that a this can be done, yeah.
00:52:17.199 --> 00:52:36.320
But sometimes a lot of business owners, especially, uh yeah, they want to clean up their infrastructure, but they fall into this trap whereby they are not actually working on things that actually benefits the company and also like yeah, benefits your own kind of uh targets that you're trying to aim for.
00:52:36.480 --> 00:52:45.519
So, in short, I think like uh like what I said, so to have uh a good context, to have someone to actually remind you.
00:52:45.599 --> 00:52:54.480
So, what are your goals that you set up right from the start, and then to keep reminding you, I think that's something that everyone should have.
00:52:54.719 --> 00:53:02.480
And that's where like having a good agent set up, um, yeah, no matter how simple it can be, I think it's gonna be very helpful.
00:53:02.639 --> 00:53:03.039
Yeah.
00:53:04.480 --> 00:53:08.639
Yeah, I I overbuild so many things all the time as well.
00:53:08.800 --> 00:53:09.840
I think it's it's very tempting.
00:53:10.159 --> 00:53:13.440
It's the form of a productive procrastination.
00:53:13.679 --> 00:53:15.119
You don't do anything, yeah.
00:53:15.280 --> 00:53:17.679
Yeah, you're just doing stuff, but you're not really doing anything.
00:53:17.920 --> 00:53:19.599
Yeah, maybe we are not the problem, guys.
00:53:19.760 --> 00:53:24.320
Maybe something is happening in the uh some kind of it's a gateway drug, you know.
00:53:24.400 --> 00:53:26.079
After you take the drug and you try to keep the.
00:53:26.480 --> 00:53:29.360
They realize it's the best way to make us pen on tokens.
00:53:29.760 --> 00:53:31.199
I can quickly walk through this.
00:53:31.280 --> 00:53:34.320
So this is uh But but this is a negative example.
00:53:34.559 --> 00:53:35.920
I mean you you framed it quite badly.
00:53:36.559 --> 00:53:42.000
Do you wanna No, I mean that that's on the surface of let's say productivity.
00:53:42.320 --> 00:53:42.559
Okay.
00:53:42.800 --> 00:53:46.800
Yeah, so so it's bad for productivity in a sense.
00:53:46.960 --> 00:53:54.480
But I think this one is useful to have, especially if yeah, yeah, you're talking about uh setting up permit in like different uh setups, right?
00:53:54.719 --> 00:54:00.400
So what this will actually give you is a whole yeah yeah, the whole system kind of overview.
00:54:00.639 --> 00:54:09.039
So how it actually gets uh your your data is like through cron jobs.
00:54:09.280 --> 00:54:15.519
So there's a health check on on every of my let's say um age agents and things like that.
00:54:15.920 --> 00:54:24.079
So it will be pinging like maybe on every hour and things like that, then it will actually appear here whenever I refresh.
00:54:24.480 --> 00:54:28.239
Yeah, so you can see the let's say my my Hermes gateway, are they healthy?
00:54:28.400 --> 00:54:30.480
How long have they been running and things?
00:54:30.719 --> 00:54:34.400
Uh and my cron jobs are they actually running?
00:54:34.639 --> 00:54:40.079
So this will be a very good, let's say, uh dashboard to find out yeah.
00:54:40.159 --> 00:54:43.599
So is did any of your cron jobs break?
00:54:43.760 --> 00:54:48.079
Is any of your agents like down and yeah, so we can see everything here.
00:54:48.320 --> 00:54:49.840
So this is updated.
00:54:50.079 --> 00:54:51.199
Uh yes, correct.
00:54:51.360 --> 00:54:51.840
Oh okay.
00:54:52.000 --> 00:54:53.519
What is the speaker thing that you had?
00:54:53.679 --> 00:54:55.519
6,000 things.
00:54:55.760 --> 00:54:56.320
Oh, okay.
00:54:56.559 --> 00:54:57.840
Wait, wait, wait, where?
00:54:58.239 --> 00:54:59.119
Up, up, up.
00:55:00.079 --> 00:55:00.639
Yeah, here.
00:55:00.880 --> 00:55:02.480
6292 runs.
00:55:04.239 --> 00:55:04.800
Speaker remote.
00:55:10.000 --> 00:55:11.760
I need to do so.
00:55:12.000 --> 00:55:14.239
This is something uh interesting.
00:55:14.320 --> 00:55:15.519
So I did a hack.
00:55:15.599 --> 00:55:17.199
So I have Google Home.
00:55:17.679 --> 00:55:21.119
So I have Google Home speakers in every of my rooms.
00:55:21.920 --> 00:55:30.719
So what I did was uh I hooked up Hermes to actually send reminders to the speakers through calendar.
00:55:31.519 --> 00:55:36.559
So for example, right, uh so my daughter will have swimming on at five o'clock.
00:55:36.880 --> 00:55:45.519
So at five uh at 4 50, the speaker will start reminding her, yeah, hey, it's time to go down for for your swimming lessons, swimming lessons.
00:55:46.079 --> 00:55:46.320
Yeah.
00:55:46.639 --> 00:55:49.039
Is it in your voice or Google Voice voice?
00:55:49.199 --> 00:55:52.639
Uh I I tried to do it in my voice, but yeah, the heck didn't work.
00:55:52.800 --> 00:55:55.519
So yeah.
00:55:56.400 --> 00:55:57.199
That's quite useful.
00:55:58.159 --> 00:55:59.760
Yeah, it's fun.
00:56:00.960 --> 00:56:01.599
Yeah, yeah.
00:56:01.760 --> 00:56:06.800
I have a home assistant, uh, and it can tell me when the bus is coming, so that's nice.
00:56:07.280 --> 00:56:07.440
Yeah.
00:56:08.079 --> 00:56:08.400
Okay.
00:56:09.840 --> 00:56:13.440
I I think yeah, back to your productive procrastination point.
00:56:13.519 --> 00:56:16.400
We have an episode coming up with Vaughn Tan.
00:56:16.800 --> 00:56:20.960
Um he he advocates a lot on like boring and tiny tools.
00:56:21.199 --> 00:56:31.199
And I think what his design principle showed at least was um that he usually limits like the amount of things he can do with his software.
00:56:31.280 --> 00:56:32.719
So it's usually quite dumb.
00:56:32.960 --> 00:56:41.280
But like he also showed his to-do list and it was actually Yeah, quite interesting, and I think it's quite effective for him to uh to keep himself on track.
00:56:41.440 --> 00:56:43.920
So for those who are interested, you can check that out.
00:56:44.079 --> 00:56:47.039
Um actually I think the rating portion is quite good.
00:56:47.519 --> 00:56:57.199
You you kind of get someone that's not emotional, attached to you to tell you that hey, this week you didn't do too well, you know, you should work better, yeah.
00:56:57.360 --> 00:56:57.920
Focus on this.
00:56:58.400 --> 00:57:03.119
So Ben, you you left Omnidesk after eight years, uh, you took a month off.
00:57:03.280 --> 00:57:23.519
Uh what's been on your mind recently, you know, in terms of uh you know new beliefs that you have with AI coming uh really fast at us and and what's next for you I think to leave omnidask after eight years is it's been like a long time coming.
00:57:23.679 --> 00:57:33.760
So especially like with the with the coming of age of AI, I think that's that's probably one of the push factors as well.
00:57:34.079 --> 00:57:44.239
So when I was in Omnidas, I think a lot of times it's like a lot of firefighting, a lot of approvals, a lot of things that um we're doing on a BAU basis.
00:57:44.480 --> 00:57:48.639
So when
AI comes about it's like wow I can start the build again.
00:57:48.960 --> 00:58:03.599
Because it's like I don't need to depend on uh agencies, I don't need to depend on developers to come up with things that I want or come up with things that uh that are very complicated.
00:58:03.920 --> 00:58:21.920
So I think that's one of the key things that uh I started to know and learn and that's where when I came out that sabbatical was more to yeah, really understand um what I want to do down the line.
00:58:22.079 --> 00:58:48.639
I think it's it's more to probably that one month is still not enough, but uh yeah, I mean it is probably one month is not enough, but I think it's more to yeah, making use of AI to start building things and as you build, I think it's more to building what you really want rather than like going according to hype.
00:58:48.800 --> 00:58:59.440
So I think that's where um my new so-called like um business that's coming up, uh I don't mind to rebuild it, so it's more towards a tea business.
00:58:59.679 --> 00:59:03.199
So it's more to bringing in premium Chinese tea into the market.
00:59:03.599 --> 00:59:12.159
So because of this AI kind of thing, uh people are just like overwhelmed with digital technology and things like that.
00:59:12.400 --> 00:59:20.079
So one of the getaways that I realized that I always wanted was more to yeah, going towards the analog lifestyle.
00:59:20.400 --> 00:59:29.679
So uh things like um photography, things like um final records, things like CDs, things like yeah music, piano behind you.
00:59:30.000 --> 00:59:31.199
Music, yes, correct.
00:59:31.360 --> 00:59:40.400
So to have tea, to have coffee and and all these kinds that takes you away from the the fast paced kind of thing.
00:59:40.559 --> 00:59:43.280
I think we talk about fast paced like previously, right?
00:59:43.440 --> 00:59:46.159
But right now with AI I think it's it's much much worse.
00:59:46.400 --> 00:59:59.039
And people don't even realize it because the the way that we're trying to catch up with AI is like it's it's crazy and we are working more than yeah what we have been doing, let's say for the past few years.
00:59:59.280 --> 01:00:03.599
And people don't don't really realize it, and that's where people get burnt out.
01:00:03.679 --> 01:00:25.280
So yeah, so I want to make use of AI, um but not like you know purely on the AI business and things like that, but more towards like building things that will actually be um yeah, useful for people, um and probably like what I have passion in as well.
01:00:25.519 --> 01:00:38.239
So yeah, I I do have first press, like first press I think is more towards like um something that I want to have to help like um personal friends like founders and things like that to get into AI.
01:00:38.400 --> 01:00:44.320
So it's more of like uh I won't say much of consultancy but more of an advisory kind of thing.
01:00:44.480 --> 01:00:54.320
So it's it's always a good to have, but at the at the
back of my head, I think I still want to have a business that I truly have passion in, and that's where the the difference comes in.
01:00:54.400 --> 01:01:01.760
So I think over that month I didn't have my laptop, so that gave me a lot of like reading room as well.
01:01:02.000 --> 01:01:04.880
So yeah, that that analog kind of thing, yeah.
01:01:04.960 --> 01:01:10.320
That I even start to read books and things like that, like during during that period of time.
01:01:10.400 --> 01:01:15.599
It's like I have never read a book for like don't know how many years.
01:01:15.760 --> 01:01:19.440
So for during that month I managed to like finish like two to three books.
01:01:19.519 --> 01:01:21.440
I mean it's it's it's amazing.
01:01:21.920 --> 01:01:26.719
Yeah, so I think that's the that's the gist of what I get, yeah.
01:01:26.800 --> 01:01:29.119
Let's say from that sabbatical, yeah.
01:01:30.239 --> 01:01:30.880
Yeah.
01:01:31.760 --> 01:01:33.280
No, that's great, that's great.
01:01:33.360 --> 01:01:40.239
And yeah, I I'm I imagine that it's the pace of life has been fast as you know, running a company.
01:01:40.400 --> 01:01:45.360
And um yeah, all the best with your tea business and first press.
01:01:47.440 --> 01:01:48.079
Yeah.
01:01:48.480 --> 01:01:49.199
Alright.
01:01:49.519 --> 01:01:52.480
So maybe yeah.
01:01:52.800 --> 01:01:54.719
Maybe we can move on to the quick firearm.
01:01:55.199 --> 01:01:55.440
Yes.
01:01:55.519 --> 01:01:57.280
So so Ben, are you ready?
01:01:58.800 --> 01:01:59.119
Okay.
01:02:01.760 --> 01:02:02.159
Alright.
01:02:02.239 --> 01:02:04.000
So what's two to three books you recommend?
01:02:04.159 --> 01:02:05.519
Two to three, yeah.
01:02:05.760 --> 01:02:08.000
Uh I don't want non-fiction.
01:02:08.159 --> 01:02:11.760
Um Rami, Kafka by the shop.
01:02:12.000 --> 01:02:13.039
Kafka on the shore.
01:02:13.199 --> 01:02:15.599
So I think that's one of the more memorable ones.
01:02:15.760 --> 01:02:17.519
Um yeah, I think probably that only.
01:02:17.679 --> 01:02:17.840
Okay.
01:02:22.559 --> 01:02:24.239
Trying to make us slow down.
01:02:25.599 --> 01:02:25.840
Okay.
01:02:26.159 --> 01:02:29.440
A favorite show, film, or content that you enjoy?
01:02:29.679 --> 01:02:33.760
Film, uh show, uh content la can be podcast.
01:02:33.920 --> 01:02:35.920
Hopefully fiction or not, not non-fiction.
01:02:36.320 --> 01:02:37.840
I would say Spider-Man.
01:02:38.000 --> 01:02:39.199
Oh, the new one?
01:02:39.360 --> 01:02:40.719
Yeah, not the new one.
01:02:40.960 --> 01:02:41.519
Or generally.
01:02:42.159 --> 01:02:43.840
The the animated version.
01:02:44.000 --> 01:02:44.719
I forgot the name.
01:02:45.840 --> 01:02:46.639
Yes, correct.
01:02:46.880 --> 01:02:47.199
Okay.
01:02:47.760 --> 01:02:53.519
So uh I wasn't much into Spider-Man, but when I watched that it was like, wow, okay.
01:02:53.599 --> 01:02:57.440
That was because my four-year-old was uh my three-year-old is into Spider-Man.
01:02:57.519 -->
01:02:59.039
So I decided to watch that with him.
01:02:59.119 --> 01:02:59.519
Yeah.
01:03:00.480 --> 01:03:01.760
Across the Spider-Verse.
01:03:04.239 --> 01:03:04.960
Yeah, yeah.
01:03:05.119 --> 01:03:07.440
Um a place that's special to you in Singapore.
01:03:07.760 --> 01:03:09.199
Juo, Chungi Airport.
01:03:09.360 --> 01:03:09.920
I love that.
01:03:10.000 --> 01:03:12.480
Um a line of philosophy you go back to.
01:03:12.719 --> 01:03:13.840
Gary Bernicek.
01:03:14.559 --> 01:03:16.960
Um you're never too old.
01:03:19.119 --> 01:03:19.519
Yeah.
01:03:19.840 --> 01:03:22.000
So you're still young to do anything.
01:03:22.159 --> 01:03:24.239
I think that's that's very relevant right now.
01:03:24.400 --> 01:03:26.639
Because yeah, I I want to do things now.
01:03:26.880 --> 01:03:32.239
And if you look at people in the room, I think yeah, so they are fresh grads and things like that.
01:03:32.320 --> 01:03:33.920
So this gives me that motivation.
01:03:34.000 --> 01:03:34.320
Yeah.
01:03:35.199 --> 01:03:35.840
Nice.
01:03:36.239 --> 01:03:38.960
A tool or product that you recently got excited about.
01:03:39.519 --> 01:03:40.239
Your camera.
01:03:41.039 --> 01:03:41.920
Favorite product.
01:03:42.159 --> 01:03:42.400
Okay.
01:03:43.599 --> 01:03:46.719
Favorite product, um, a pint of Guinness.
01:03:49.920 --> 01:03:54.400
So it because this is a podcast, I don't want it to sound I don't want to sound like alcoholic.
01:03:54.719 --> 01:03:56.719
So why guineas?
01:03:56.800 --> 01:04:05.679
Is because I think there are a lot of situations whereby um a lot of people don't, or rather, not many people, okay.
01:04:05.760 --> 01:04:10.639
Yeah, go go for guinea, uh, go for go for whiskey and things like that to slow down.
01:04:10.880 --> 01:04:14.559
Um I like whiskey, so but not a lot of people do that.
01:04:14.719 --> 01:04:18.800
Um I don't drink beer, so I only drink guinness.
01:04:19.119 --> 01:04:31.599
And by changing that, right, so I start drinking guineas, there are a lot of conversations with a lot of different people that yeah, I made a lot of good connections there, and it was very, very worthwhile.
01:04:31.679 --> 01:04:36.719
And the conversations that I had over those drinks, I think, yeah, made a lot of difference in my life.
01:04:36.800 --> 01:04:39.280
Yeah, very valuable drink.
01:04:40.639 --> 01:04:41.119
Nice.
01:04:41.280 --> 01:04:49.119
Alright, the last one, uh, where can people find you, you know, talk to you and engage with your work and how can they be helpful to you?
01:04:50.000 --> 01:04:57.679
So, right now, um, very much on uh you can contact me via LinkedIn um or first press.ai.
01:04:58.079 --> 01:05:19.760
So basically, if there's uh if you're a DTC founder, um I won't say a starting founder, but an existing founder who has at least like uh two to five mil kind of thing, and you're looking to um to improve on your business using AI, and you have there are no place to start, uh just talk to me.
01:05:19.840 --> 01:05:21.199
I'm open for conversations.
01:05:21.360 --> 01:05:24.239
So I think that's one of the key things that I like.
01:05:24.400 --> 01:05:32.159
So because I'm doing the solo founder kind of uh situation now, so I really like to go out and talk to people.
01:05:32.320 --> 01:05:38.960
I mean we still can have Google Meet and things like that, but yeah, to have a coffee or even a kind of guineas, I think that's great.
01:05:39.119 --> 01:05:41.119
So I'm looking forward to this kind of talks.
01:05:41.519 --> 01:05:42.159
Yeah.
01:05:42.719 --> 01:05:45.360
Then um, what else?
01:05:46.559 --> 01:05:47.440
Okay, alright.
01:05:47.519 --> 01:05:48.559
Thanks so much, Ben.
01:05:48.719 --> 01:05:51.119
Thanks for joining us and sharing so openly as well.
01:05:51.280 --> 01:05:52.239
Really appreciate it.
01:05:52.480 --> 01:05:53.599
Okay, thank you again.
01:05:53.760 --> 01:05:55.280
All right, all right, bye.
01:05:56.159 --> 01:05:58.719
Thank you so much for tuning in to the Code Riff podcast.
01:05:58.880 --> 01:06:01.519
We hope you learned and enjoyed the conversation as much as we did.
01:06:01.760 --> 01:06:07.440
If you'd like to support us, please consider subscribing, leaving a comment, like, or review on wherever you listen to your podcast.
01:06:07.679 --> 01:06:09.440
And we hope to see you in the next episode.
01:06:09.679 --> 01:06:10.159
Bye.