Stephen x Can Weekly Sync — Full Transcript

Source: Telegram DOCX attachment (file_121---090cbfa1-1a15-4536-80fc-8d0d0a39ad1e.docx) Date in transcript context: 2026-03-06

Stephen & Can — Weekly Sync

Meeting Transcript — Cleaned

Executive Summary

Topic 1: PoutineAI — Invoice Processing App for Quebec Restaurants

The app is operational and ready for initial demos. The email-based invoice extraction feature is working and represents the strongest demo hook. WhatsApp integration is currently blocked due to a Twilio/Meta issue — Stephen has raised a support ticket. The team agreed to launch without WhatsApp rather than delay further. Can will begin outreach to 25–30 restaurant customers in Quebec for initial demos and feedback. Content strategy will focus on short (30-second) Instagram videos demonstrating the invoice management features, with Can producing the content and Stephen handling digital distribution and repurposing. A simple placeholder logo (black and yellow) will be produced quickly to allow Instagram launch, with the full branded logo to follow. Payments integration is deferred — Can will reach out to Vopay for pre-authorised debits/EFT processing when the time is right. Stephen shared the credit card cashback research showing that the effective cost of card payments for restaurants may be lower than expected (~0.6% net after cashback). The team discussed a per-customer agent framework (inspired by a Shopify friend’s architecture) for capturing user feedback, intent, and bugs automatically.

Topic 2: New Software Opportunity — Wholesale Sales Platform (Toilet Paper / Cash-and-Carry)

Can introduced a paid software opportunity from a high school friend who owns part of a large Turkish conglomerate (Hayat Kimya) that is investing heavily in UK paper manufacturing. The group has acquired multiple UK factories (Nats, Modern Paper, Diamond Box) and is building Shotton Mill — the largest paper factory in Europe (£1.5B project). The factories currently operate in silos with different back-end systems (Sage, Index, Excel) and compete against each other for the same customers. The friend wants a unified WhatsApp-based sales and ordering platform for approximately 250 cash-and-carry customers (pilot in one factory). Core requirements include: WhatsApp-based ordering with catalogue browsing, instant confirmation, and notifications for special deals with deadlines; payment options including credit card (pay now) and 30-day invoicing with Sage integration; order tracking (shipped/ETA notifications); data analytics on customer responsiveness and offer performance with weekly sales reports; and elimination of manual email and phone-based ordering. Stephen recommended a WhatsApp-first approach rather than a native iOS app, using agents for ordering, customer support, and analytics. Payments could be handled via Stripe for credit cards, with UK bank transfer APIs for larger transactions. Stephen will provide tiered pricing (skeleton MVP first, then add-ons). Can will send a written list of core features and get pricing approved by the owner. The pilot would start with one factory and ~200 customers.

Action Items

Stephen’s Actions

PoutineAI

Continue bug fixing and automated testing with AI agents (English/French switching, UI regressions)

Follow up on Twilio support ticket to resolve WhatsApp number connectivity issue

Restore the phone number input field in the WhatsApp feature

Build the content repurposing/delivery engine for distributing Can’s video content

Set up YouTube channel for PoutineAI (or manage YouTube uploads)

Generate placeholder logos via Nano Banana API using Friday agent for Can’s review

Update the PoutineAI site placeholder logo and branding once Can provides the simple logo

Run go-to-market research for PoutineAI (same approach as OYN research)

Begin building the per-customer agent framework for feedback capture (backlog — after bugs are resolved)

Prepare and send meeting summary with sprint actions for the coming week

New Opportunity (Wholesale Platform)

Prepare tiered pricing proposal: skeleton MVP price and add-on feature pricing

Review Can’s core feature list when received and scope the technical architecture

Research UK bank payment APIs (Open Banking / Faster Payments) for high-value B2B transactions

Consider Sage integration approach for automated invoicing

Can’s Actions

PoutineAI

Test the email invoice extraction feature one final time to confirm it’s working

Begin outreach to 25–30 restaurant customers in Quebec for initial demos

Produce short demo videos (30 seconds to 5 minutes) showing invoice management features

Follow the video script structure: pain point → credibility (“why listen to me”) → solution → call to action

Create the PoutineAI Instagram account (even with placeholder logo)

Upload video content to shared Google Drive or send Loom links to Stephen for repurposing

Commit to producing a couple of videos per week

Use Loom to report any bugs found during demos (quick screen recordings sent to Stephen)

Get the simple black-and-yellow logo from the designer and share with Stephen

Reach out to Vopay to explore pre-authorised debit/EFT payment integration (when timing is right)

Ask for user-generated content from early users (e.g., a short Instagram post or video in exchange for free access)

New Opportunity (Wholesale Platform)

Write up and send Stephen the list of core features for the wholesale sales platform

Confirm what CRM/back-end systems are currently in use across the factories (likely Excel + Sage + Index)

Find out where the factory is losing money (to help scope value-based pricing)

Present Stephen’s pricing proposal to the factory owner for approval

Advocate for WhatsApp-first approach over native iOS app with the owner

Confirm pilot scope: one factory, ~200 customers

Full Cleaned Transcript

Stephen: Hello.

Can: Hi Stephen, how are you?

Stephen: Good. I just want to make sure the audio is recording. I just realised that my Google account doesn't have recording privileges, but I don't care. I just record the audio with my mini recorder so it's easier that way. And if we need to do visual, I'll just get you to do a Loom or something.

Can: I also couldn't do it. It says recording unavailable on my end as well. Is it because you don't have permission or I don't have permission?

Stephen: It's just that I have the basic plan and I don't feel like paying for the other one. I do have Teams with my service business, but I don't want to pollute that with recordings and stuff. So it's better to keep it separate.

Can: All right. Life is good?

Stephen: Life is good, yeah. Just busy. Trying to push to the final mile for my two products — yours and then the financial one.

Can: How's the other one doing?

Stephen: It's great. I did a lot of research on go-to-market and I think it has a lot of possibility. Right now I'm trying to package it up and apply to the Xero App Store. There's really nobody doing anything comparative. The only comparative one was bought by Sage. I'm just trying to push. My partner just doesn't move as fast. He's a little bit more old school, plus he has all his financial clients. So I'm like, we've got to get going now. We've got the landing page, we've got that stuff. He's making content, which is good. So I'm now working on a repurposing engine, trying to find the right one to take the content and distribute it across our main places, which are mostly LinkedIn, but also YouTube. I did a lot of work on keyword searches. My agent is amazing — she just works all night and finds stuff. I've scraped all the App Store stuff. I've done the same thing for Poutine as well. I have a webpage. I don't know if you've gone to it.

Can: The Poutine research?

Stephen: Yeah, did you see the webpage? There's a web page that she makes. Let me share my screen.

Can: Sorry, no, I didn't see it. Did I get an email link?

Stephen: It's not an email link. I think it was a Pingju or something. Let me just find it. It would be the Agent Maker one — that's the one I sent you for Paraguay.

Can: Yeah, I do have access to that — the payment and the Paraguay stuff.

Stephen: Yeah, so she did it for payments and stuff. That's the next step, but I haven't done the go-to-market one yet. I've done that for the other product. Let me show you.

Can: Yes, this one I checked.

Stephen: OK so I have to update the go-to-market research, but I've done this — it's basically scraped all of Xero, all of Reddit, found all the pain points, looked for where the best channels are for this marketplace. About 80% of it is on the Xero marketplace. It even made guidance for my partner — like LinkedIn posting, step-by-step. Then I've got a whole bunch of stuff on SEO, the keyword report on our domain. Nobody's really even advertising. They just put it on the marketplace. And then also looking at LinkedIn Sales Navigator, because for our market it's different. You said yours is mostly on Instagram and Facebook.

Can: Yep. Non-digital, yeah. I'd say mostly 90% Instagram. Maybe 10–15% Facebook. But 90% Instagram. That's what I'll go for.

Stephen: Yeah. For the marketing bit, I can do go-to-market engineering. I'm learning about it, I'm getting there. I know the distribution — we even monitor forums. One of the things we came up with is answering questions on Xero forums and things like that. We're going to set up a School community where we can do open office hours to teach people. Anyway, we've got that, and that's sort of what I want to do for Poutine once we get it going. I have to do the same type of research on go-to-market for Poutine, and I'll let you do the analogue stuff.

Can: Yeah, the videos — I'll do the videos.

Stephen: Yeah, because you can be the trusted face. You do the videos and I'll do the digital marketing with the content you create. We can work on it together, obviously I'll pass it by you. But I'm building a content delivery engine. We just need to think about what type of content would be interesting. They're busy — probably the restaurant owners — so it's not the same type of thing. It'll be a different type of strategy.

Can: We'll do very short videos, yeah, like 30-second videos. Just to show the invoice management aspect of it. I do have some ideas. I've been talking to some people who still live in Montreal. I'll do some videos. I'll also ask their help in the restaurant community to do some videos. I think it's going to be good. A lot from Instagram.

Stephen: I can scrape all that. Do you have an Instagram account already?

Can: Personal, but not for Poutine.

Stephen: For Poutine, do you have one?

Can: No, I'm waiting for a logo first. I want to launch with a proper logo. I ordered a nice logo from a friend. Once we have a logo, we'll start the Instagram page. I'll slowly start putting up some teasers — countdowns, "get ready to save time on your invoices" kind of content. But first the logo. I want to share the logo first with the branding and something like "coming soon."

Stephen: OK, cool. So like, you can see my agent — ChatGPT just upgraded again and it's even more brutally good. It's doing end-to-end validation and testing right now because it found a bunch of bugs. This is GPT-4 that just released last night or this morning. I installed it and it just keeps going. It's amazing. It's even better.

Can: I thought Gemini answers were getting much better.

Stephen: Yeah, for my side it's not as good at code and it over-engineers, but I do use it for deep research. For things like generating content — did I send you that video on go-to-market engineering?

Can: The marketing world one?

Stephen: Yeah, that's brutal. You can see how you can do A/B testing easily, especially with long-tail ads on Instagram or Facebook or wherever restaurants hang out. There's a whole bunch of go-to-market engineering we can do and play around with. I'm trying to build a software factory now with these agents. I've got three of them running right now on different projects plus the one at Cisco. We can sort of build a marketing engine, like that guy did in the video. You can do a lot of A/B testing, and once it's running, I can combine it into an agent that runs all the time. You can see testing, see how many clicks you're getting, take down bad ads, optimise for good ads. You can look at what other people are doing, what's going viral, look at what your competitors are doing and pretty much copy them. There's a lot you can do. You can also do keyword SEO searches like "alternative to" whatever they currently use in Canada.

Can: Well, maybe Pluto a little bit.

Stephen: Yeah.

Can: So your partner's name is Hamish?

Stephen: No, I have two other partners. One is for the services business. We just do things for companies. I've got a client — it's a pharma agency. They do research. I'm basically building a research agent for them, which is what I've already built for myself. It just automates — they've got a bunch of agents they're going to talk to in Teams. It does competitive research and organises all the stuff in SharePoint. That's what they do manually now, so I'm just automating it. The other one is the financial direction platform.

Can: The online CTO/CFO?

Stephen: This is it. We've already launched it. We have a waitlist, although my partner isn't pushing it as much as I'd have liked. This is the app.

Can: That's Hamish, right? I see his LinkedIn posts sometimes.

Stephen: Hamish — yeah, that's the service one. Ciaran is the other guy. So this is our thing and we actually do have a waitlist. Right now you can upload Excel exports from Xero. It's called Own Your Numbers. Let me show you the website. Hopefully it's working because the guy who wrote the website is from the Isle of Wight. Oh, he's screwed it up again. The caching is broken. I've got to send this to him — I'm not happy about this. I could write a better website. How are we supposed to get people on our waitlist if it's down? It's supposed to go to the waitlist here. You put in your email. It's very slick actually. And then there's a basic Loom video.

[Loom video plays]: "Nice to see you. Welcome to our website."

Stephen: He runs through different types of numbers. Essentially it gives you a cash score based on his own methodology. And then it gives some AI-driven recommendations on what to do. After this, you just watch the video.

Can: Is this one free now?

Stephen: It's pretty much free now. It costs me like one cent to process. You just upload your Xero exports, although what I'm trying to do now is the new part where you connect directly to Xero. The data stays secure — it's zero-knowledge. I just pull the data from Xero, we run the calculations, give the report, and it takes about thirty seconds. Ciaran charges about a thousand pounds per month and it takes him forty-five minutes — and he's an Excel god. So that's the type of value. Now I'm applying to put it in the Xero App Store. But the problem is he went on a retreat for a week. I'd just built the Xero connection and he was testing it. He proxied some of his clients to me to test while he was away, but you can't do that from inside Xero. So I had to build into the application that he connects his clients because he's a financial director with lots of clients, and then inside the app he proxies to me. He proxied it to my Xero, but I can't actually see it. I found this out on Monday and he'd already gone off. So I'm just waiting. As soon as he does it, I'm sending the package to Xero. I think we also have potential to put it on QuickBooks or something.

Can: I think that's the best next step. That being said, most of these guys don't even have QuickBooks. They just bring a bunch of receipts to their accountants at the end of the month. So when they do a software purchase, it's not going to be through QuickBooks — it's going to be through Instagram, honestly.

Stephen: Yeah.

Can: For bigger restaurants at some point, we'll need to.

Stephen: Yeah. So like I said, I've been going through it, fixing bugs, the agent's working on them now. I found a couple things between switching between English and French where they didn't have the titles right, but it seems more or less OK now. The WhatsApp one — I added some stuff. It regressed. It used to have the phone number field to send to. I have to put that back in. We haven't got WhatsApp working right now — I don't know if Meta shut it down because we weren't using it. I have a support case open with Twilio. I put it as critical. But the problem is the numbers are showing rejected, but on the same side it's showing passed. It makes me crazy because there's no way of validating it. It says this one — name visible to customers — Poutine AI was approved, but then it says it's pending. And then the one they said they approved the other day has now been rejected. The emails say different things from what the console shows. If I go to my WhatsApp numbers, for whatever reason they're offline. It's sort of like Twilio integrating with WhatsApp, but WhatsApp says they're valid. According to the AI and everything, this pending thing is just for the display name rather than just a phone number. I don't know why they disconnected it because those numbers used to work.

Can: Maybe some of those are my attempts.

Stephen: Oh maybe. This is me doing some testing. This is outgoing. But it was working — we could see the messages.

Can: It was working. We did verify it. It was working from my Canadian number.

Stephen: Yeah, I was testing it and it does work because these are verified. I went through the process and verified them again, had it call back to my phone number, put in the code. They're all verified. It's just the connection between — I think it's on Twilio's side. Hopefully they can help me. I did the verification yesterday and they were fine because I could call the number and it came to my UK phone number and I got the code. So I know they're working. It just has to do with WhatsApp integration into Twilio. Anyway, it's annoying.

Can: Do they have a contact number or not really?

Stephen: I raised a ticket. I'll just try to keep following up and keep pushing them.

Can: Listen, we can totally launch it without the WhatsApp.

Stephen: I think it might be a nice thing that if you do enough content, there's enough things here — maybe it's good to go through it and start making Loom stuff. It doesn't have to be fancy. You saw my colleague's ones, right? It was just walking through the app and telling a nice story.

[Loom video plays]: "Who owes us money?"

Stephen: He explained how it worked and showed the cash position. Just do a few of them, even if they're short, and we can put them up on Instagram, YouTube, and work from there. Even if you show little segments. The main one is just talk about the overall story. You know, here's the pain. Who am I? Why should you listen to me? Because you've got a lot of restaurant experience. This is the pain — your mum's doing it, you're doing it on weekends, it's costing you $500, whatever. Just talk it up. That's how all the YouTube scripts work because I've studied them a lot. Second part: why should you listen to me? Quick thing about — I'm Can, I know the import wine business, blah blah blah. Then: this is the solution, this is how I'm going to help ease your pain. Sign up here or whatever we're going to do. We need to find the call to action after you demo it. So if you want this, get on the waitlist, or we have an email address, however we want to handle it.

Can: Yeah. I'll do one.

Stephen: If you do a bunch of them showing the different features.

Can: Maybe I'll email it to people first. I'll just email it to some customers — 25 to 30 of them — to get an initial reaction.

Stephen: There's also — even if you don't do it through Loom, like Ciaran, if I show you his LinkedIn. This is also what he's doing because most of our audience is on LinkedIn. He just did one based on Federer — he tells a story as well. He's very good at it. He has some software where he can switch between PowerPoint view and then he talks about the app and storytells a bit about why cash is important. He's really good at it.

Can: He tells a story and then he shows the product?

Stephen: Yeah, I don't know exactly what software he uses. He's on Windows. He tells a story and then I'm getting him to do actual demo content. When he has it done, he uploads it — he uses Dropbox. We can use a shared Google Drive. Just upload it there and I can start repurposing it. That's all automation — that's what I do. So if you focus on content and the analogue side, and just keep uploading content. Commit to doing a couple of videos per week and a bunch of outreach with that content, whether it's Loom or whatever. Loom is easy because you can just share a link in an email. Even if you can't download the Loom link, doesn't matter. Give me the Loom link — I have a downloader app that I made with AI that can grab them. Don't pay for the extra one, just do the basic Loom, even the free one. I think the free one lets you do five minutes.

Can: Five minutes.

Stephen: Five minutes — and you probably don't want more than five minutes, right?

Can: Yeah, I'll expense it to my company and we'll get to use it a lot. I now enjoy using it, so I'll get the professional version.

Stephen: Yeah, and it's very good for giving me feedback too. When you're demoing, if you find something broken, just do a quick Loom, send it to me, and I'll fix it. We'll try to iterate fast like that.

Can: Yeah. So your research suggested Vopay. Shall I get in touch with them? Vopay is the sales platform that facilitates bank payments in Canada. The research also suggested Pluto, but Pluto is a competitor — a direct competitor. So this is for the pre-authorised debits, EFT payments, because technically we still can't process them. We need some compliance that's going to be very expensive. Your research suggested Vopay and I checked them out. They look OK. Should I get in touch with them?

Stephen: Sure, sure, sure.

Can: We can do that in a few months before we integrate sales. I think we should launch without sales, without waiting longer.

Stephen: No, no, we do it without — yeah. Even if it's just free. It costs me nothing to run this. It costs me like one-tenth of one-hundredth of a cent.

Can: After a few months, let's see people using it. They're going to generate some content too. We could give it to some people and just ask for a quick video from them as a payment. Just to say thank you. We'll give it to you for free — just make a 10-second video about it. Or post it on your Instagram.

Stephen: That's going to be even more valuable. Totally. And the other thing too — if you can spend some time showing them around it, getting feedback. One of the things I want to do, and one of my friends does it for that Shopify friend of mine — I put it in the backlog but I'm trying to get the bugs out first — is I'm trying to build an agent framework. I've got a really good agent harness now that I'm using for everything. My friend, he's got that app in the Shopify App Store that does all the quizzes. What he's building now is almost like an AI that helps people create the quiz. But the cool thing is he's building an architecture where every single customer has their own custom agent. It's isolated, it knows them, it's their own custom agent. Not only does he use that for helping people use the tool — because it knows the codebase and how to use the tool — he also uses it to collect feedback. When people say "how do you do this" or "can you do this" and it's not available, he starts adding those as features. The AI collects it, puts it in GitHub, and adds the features. Or if it realises people are struggling or saying there's a bug, it automatically goes and starts looking at the codebase and finding bugs. So it's a way of catching intent and what people want to try to do, even if you don't have that feature set.

Can: Yeah, that's what we talked about.

Stephen: Yeah, it's amazing. But he's trying to do it with one agent per client. He said he'll share his architecture. So I'm going to build something like that — almost like a customer service chatbot. Some people like them, some don't, but for us it will help capture things. And then on the side of that, when we see somebody engaging with it, if we want to jump in as founders while we're ramping up, one of the things he finds amazing is that people love it when — if they're struggling or whatever, he's got this functionality in the back where he can jump right into the customer service session. He can see what they're doing. And they're like, "Oh my God, it's the founder!" I won't be able to do it in French though, because I'm not fluent, but maybe you can help, or I can use translation.

Can: Yeah, honestly most of the guys I interact with speak English. I normally speak English to them. Sometimes I speak French, but most of the time we speak English.

Stephen: OK, well you can do the content in French or English. Both, actually. It wouldn't hurt to put both on the page — just more content.

Can: Yeah.

Stephen: Yeah.

Can: Is the email feature working?

Stephen: It should. It should work fine.

Can: I think the invoice feature works. The product is operational. I can start demoing with the product. It doesn't have to be perfect. The extractions can come after. For me, the magic part is really getting the invoices from the emails. That's very impressive for people. So let me start showing it to people. After that, next week we can have WhatsApp maybe, but let me start showing it to some people. Let me do some videos in the meantime. And then I'd say we have a working product after that.

Stephen: Yeah, and if you can start looking at content. As soon as you get content, I can start farming it out. If you can also open an Instagram — and I can do the YouTube if you want. If you want, wait for the logo, but you can still open the Instagram and then add the logo to it later. If you have some manual time when you're just killing time at the factory, if you can help me with some of those tasks, I'd appreciate it.

Can: The designer is going to do a quick logo with the name first. Black and yellow. So we can start launching with that before we get the real logo with a mascot and everything.

Stephen: Yeah, and as soon as you get it, I can replace the placeholder on the site and update the branding. So I'll give you the summary of this, we'll all have our actions, I'll run it through the AI, and this will give us the next week's sprint.

Can: OK. Stephen, would you like me to tell you about what my factory friend is looking for?

Stephen: Yes, I would love it.

Can: So we sell toilet paper, right? Many of his customers are cash-and-carries. Do you know the cash-and-carries?

Stephen: Yep.

Can: Most of them are family-owned. Most of them are still old school. "Call me and I'll get your order" or a lot of back and forth on email or on WhatsApp. He's not a tech guy. But he's thinking — and these guys are not tech guys either — he wants something where he can send notifications to these cash-and-carries with some special deals. And obviously he wants to get rid of that push-and-pull. Let me see my notes. He's looking for an iOS app. I said no one uses apps anymore — it's all agents and WhatsApp.

Stephen: It's all like wherever their people live, and mostly it's WhatsApp. Yeah.

Can: Exactly. So probably WhatsApp is going to be a better option than doing a separate app. His main thing is notifications — I said we can totally do it by WhatsApp. He wants this because right now they're sending emails or doing sales calls, and it's not working well. No one replies to the emails. Sales calls are very expensive when you think about it.

Stephen: Yeah.

Can: He wants to send special deals like "buy 10, get two free" with a deadline. He also wants an option to see the whole catalogue, and customers can add stuff to their cart from there. He also wants something where they can pay right away, with different options — maybe pay now with a credit card, or a direct bank transfer. He wants an option to pay in 30 days, because normally these people have credit with us. But he wants to do a small discount for payments made right away. He also wants to analyse the data from these transactions — which guys are more responsive to offers, which ones respond more to instant offers, so he can do more targeted offers in the future. He wants to eliminate the back and forth with email. He's getting a lot of "where's my order" calls. So maybe some order tracking. Not with GPS on trucks, but something like "it's shipped, here's your ETA." Something simple. And he wants to sell with clicking, not with "here's your invoice, here's my email." He wants to eliminate that old-school transaction — buy now, done, we'll send it in two days. And he wants to do more online sales. He wants something to show to Tesco, to Sainsbury's — "yes, we sell toilet paper, but we are a tech-heavy company. Look at the web-based app we developed for the cash-and-carries we service." And he has a budget for that. He owns the company, so he does have money. So I said, OK, I know a guy. Let me talk to him. He does want an iOS app, but I said let me talk to my guy first. Let's see what he recommends.

Stephen: Yeah, you don't need to do the app. There are a lot of parts to that. One is automating your sales team. One is instant payments, which you can even do with something like a Stripe link — a Stripe invoice link that routes to a web page where you put in your credit card details. What I'd need to know is what does he currently use for his CRM, for his client management back-end?

Can: I'm curious what he uses as well. What he uses for an order management platform.

Stephen: Well, whatever he uses for customer management. Does he use anything, or is it handwritten? The answer is probably Excel.

Can: OK, let me tell you about this company. This is a huge Turkish conglomerate. I know this guy from high school — he's a friend of mine.

Stephen: Yeah.

Can: It's a multi-billion dollar company actually. They have a huge IT team in Turkey — at least fifty people. This guy, they decided to invest heavily in the UK a few years ago. In the north of the UK, they're building the largest paper factory in all of Europe. One and a half billion pound deal. At the same time, he's buying smaller factories in the UK to first eliminate competition and second have some local presence. That's why I'm in Leicester sometimes — to kind of manage one of these factories because they bought it but got rid of the owners and kept all the management, but there's no communication with head office. They can't bring together all these factories they bought. They bought them too fast. So all these factories are using their own systems — one uses Sage, another uses a different management platform. They're not coming together. And there's this huge factory still being built in Chester, in Wales — Google Shotton Mill. It's impressive. I heard it's now the third-largest construction project in all of Europe.

Stephen: Wow.

Can: But that's why the sales system is still super underwhelming — they didn't come together.

Stephen: They grew too fast, yeah, and in silos.

Can: And in separation. So every factory is doing their own thing. He wants something that brings together all these factories, at least on the sales side. My factory is Nuts, another factory is Modern Paper, the big factory is called Shotton Mill. There's a cardboard factory called Diamond Box. But technically the customers are the same, and sometimes the factories are competing with each other for the same customer, even though they're owned by the same parent company. That's why he wants something that brings all these factories together, at least in the sales aspect. And he doesn't want to involve the IT team in Turkey — it's probably not their speciality, and he wants to show he can do things without the conglomerate's help.

Stephen: So this is exactly the area where you can play, especially if you're doing a greenfield. Going from Excel or paper or whatever old CRM, or even from something like Salesforce where they're probably paying too much and wouldn't use something like that — which is probably a good thing. This is totally the right time. You could even build your own CRM or layer it on something like Go High Level or HubSpot. You can even build your own thing if it's not that advanced. It would be interesting to know how they manage their clients and how many they have. But you can just start bolting on automations — building your digital workforce around that. Having an agent doing outbound calling, one handling inbound calls, one that's available 24 hours. When they want to put in an order, they just talk to this thing and it puts in the order for them. There's so much more. That's even just the voice one. But even the basic thing — being able to reach your customers through WhatsApp, which is less friction than going to a website, calling sales, going to some app. If you can reduce that friction and be where your customers are, you can do all that stuff — upselling, analytics, all of it can be bolted on. The cost of software is going to zero, but the best thing is you can make it custom. The cost of software development is zero. The cost of somebody like me who knows how to do that is not zero. This is what I'm trying to get through to people. I've got so many people who always ask me to teach them this and that, and I've got to start getting paid for it. Anyway, long story short — yes, I can do stuff like this. It's good that he has money and wants to be forward-thinking, and it's good to know the founder directly. Not some big conglomerate where you have to talk to fifty million decision-makers. I hate that.

Can: Literally, he's the owner. And he makes the decisions.

Stephen: Yeah.

Can: There's not going to be a layer. If you consider me a layer, you and I will probably work together on this. But it's definitely going to be paid. Let me know what you have in mind. Let me know what we can come up with. I'll obviously have to get him to approve the price, but after that, you and I will work together as if it's our own project. And yeah, we can maybe do the call agents, the ElevenLabs stuff.

Stephen: Yeah, we can do all that. It's all about agents now. They are capable — as of Christmas, they are more capable than before. My agent Friday, she's amazing. She can see across my whole world, my whole life, all these things I've got going on. Imagine having that same context and intelligence across your business. And even giving him insight. As he runs around with his WhatsApp or whatever — that's what we're trying to do with this financial thing with my other partner. It's like a busy CEO on the road doesn't want an app, doesn't want to log into a dashboard. They just want to know if they're going to run out of money next week. These high-end startups want to know their burn rate. They want to know who owes them money, who they owe. Quickly. Thirty seconds.

Can: Using an agent. I think that makes perfect sense. No need for an additional app, and we can still analyse the data. They're basically just going to write "OK, send me this" — like just normal talk, right?

Stephen: Especially if they have an account set up. They might even have something where they have to put in a password for security. If they're an existing customer on credit, they already have an account. If they're new, they set one up. But they can even have it so they're on the fly. There are so many use cases and demos — we'd have to build the core and then bolt on all these features. Automate the sales, automate customer support — all that can be automated now.

Can: If you could give me a tiered pricing — like a price for the bare minimum skeleton. It's got to involve payments.

Stephen: Payments is fine. If he's happy with Stripe, Stripe can pretty much handle everything. The problem we've got with Poutine is that Stripe really isn't the right fit because of the cost. But I don't know if that's a problem for him or not.

Can: Maybe it is a problem because these are wholesalers. Usually bigger volumes, smaller number of transactions. They're going to order like a truckload. That's going to be a £30k payment. So they'd probably want it as a bank payment rather than a credit card payment with 3%.

Stephen: Yeah, because Stripe takes about two and a half percent. But there are workarounds. If they have a direct debit, it's actually easier in the UK because the bank payment system is faster. They have APIs for it, it costs nothing, and it's instant. Whereas the US is terrible with ACH. They're bringing in FedPay now. And Canada, as you know, is totally behind with the Interac 150. They're bringing in that other thing in the third quarter. The cool thing going back to Poutine is I can start building for that because they have sandbox and early adopters, even though they haven't gone live yet. I can still build the infrastructure so as soon as they give us early access, we can start using it. Maybe they don't care about small Poutine AI companies, but at least we're ready when they flip the switch. And as you saw, it's like 15 cents or something for a payment.

Can: Yeah, very cheap. At the moment I'm paying $1.50 flat fee. For personal it's free. For business it's $1.50.

Stephen: When I do a bank transfer they still charge me $1.50 with my Canadian bank. It's not terrible. My mum sends me money and she has to pay $1.50.

Can: 3% is harsh. It gets harsh when the amounts are bigger. But your research agent actually found some interesting stuff on credit cards too — there's a cashback option that most restaurants use. The research says most cashbacks are around 2%. I never got a 2% cashback, so that was very surprising. So if the cashback is 2% and the credit card charge is 2.6%, you're only paying 0.6%, which is not terrible. I think the calculation was it makes a few thousand a year and you're getting credit for it. So you're paying for the rolling credit.

Stephen: Yeah, it's like rolling credit.

Can: OK for the restaurants, yeah.

Stephen: Yeah, and especially if — but then again you don't want to carry the balance because then it's like 29% or something ridiculous. But yeah, that's what the research found too. I was pushing back on my research agent, saying Can told me people use a lot of credit cards. They were saying the only way it makes sense is if they have enough cashback so that the carry charge is less than the actual interest, so it helps them smooth over. Because I thought — are they really paying 2.5% on every transaction with credit card? Bank transfer looks cheaper. Anyway, that was the research. But checks are also still a thing.

Can: Checks are a real payment option still in Canada.

Stephen: Which we can't really help with, I don't think. But we'll see.

Can: Anyway, for this — maybe we can do credit card payment with fees, 3% fees, pay-now option with credit card only. And then the rest of the options — these guys already have credit with us. So we can just send them an invoice. We can find a way to automate the invoices. For one of the factories we use Sage. For one we use Index. It's very old. But happy to switch that factory to Sage also if it helps us come together. So we can start looking into integration with Sage and at least get the invoices automatically, because right now someone is manually sending them. So they can order from WhatsApp, no app. They can order from WhatsApp, get instant confirmation, we send them an invoice by WhatsApp or email or both. They pay in 30 days — we'll still chase them manually for now. And something that analyses the data for us. Maybe at the end of every week we get a sales report — see what works, which offers worked, etc. And tracking. We say, "Your order has been shipped, your ETA is this." Very simple.

Stephen: Updated tracking. Yep.

Can: And yeah, then we can start adding more stuff — like "do you want to order samples?" We can maybe integrate calls. Something that auto-responds to calls. But if you can give me a price for the skeleton, I can get it approved and we can start working on it. And then I'll say OK, additional stuff is going to be this much. But he wants something to start with. He hasn't done anything like this before. He doesn't know how much software costs. The factory is losing money at the moment, so he's a bit stressed about money. Again, the money is borrowed from the family. But we are talking about a large operation.

Stephen: And I guess the thing to know is also where is he losing money? Is it on human resources? Because there are a lot of things we can attack to actually improve things. Sometimes it's hard to know where they're bleeding unless you have a good view — that's why my partner's Own Your Numbers thing can tell the person immediately where they're bleeding. So it would be good to know that. It would also be good to know — even if you just do a voice message and send it to me — what you think the core features should be. For me, it would be useful to know if they're using anything in the back end for a CRM. And also to get a scope of how many clients they have. Or if you're going to pilot this in one factory first with a certain amount of customers, and then we roll it out across and try to unify everything. Maybe it's better to do a pilot first. But it would be good to know roughly how many clients. Not because I want to price on that — that's not how I price, I price on value — but I need to know the scale to think about the right type of infrastructure.

Can: I can answer that right now. At the moment they have about 250 cash-and-carry customers. But they kind of launched about two years ago, so they're looking to grow that number. At the moment it's around 250.

Stephen: OK, cool.

Can: The CRM — I suspect the answer is going to be Excel.

Stephen: Excel. That's good. Easy to replace Excel. Even if we build something ourselves — just our own CRM. That's cool.

Can: But I'll write you the core features. Writing is easier. Give me a price. Let me talk to him. And then we can start working on it. We'll see if he insists on "no, I want an app, app, app." I hope he doesn't say that. I'll try to show the value of not having something additional — instead using what people already have, like WhatsApp, which literally all our customers probably have.

Stephen: Yeah, and where they work. That's probably the easiest. But an app is more development, more cost. I haven't done it before — it's not impossible, AI can do it, but it's just going to be more expensive as initial friction.

Can: Would you like me to come up with something with vibe coding? I told him it's not going to be great, but I could probably learn it in a month and do something. Or would you like something a bit more professional from the start? He's like, let's get some prices.

Stephen: Well, the main thing is also the security, the scalability — stuff that even my other partner, Hamish, he does some amazing stuff and spins it out and shows customers immediately. Like yesterday he was doing something for an engineering company that makes carbon fibre tubes. He basically built them an app that did 80% of what their engineering team does. And he's an ex-CEO of an ad agency — he really knows how to use AI. But that last mile of making it enterprise-ready, supportable — all the stuff I've done for Cisco. I've got stuff running that's been there 15 years. I don't even remember how it runs or where it runs — it just keeps running. That type of stuff is important, especially if you start scaling past one factory to multiple factories. Somebody's got to support it at the end of the day and make sure it's reliable.

Can: So for now, maybe assume we start with one. Imagine it's a small to medium-sized business, even though it's an enterprise at the end of the day. He's probably going to want to test it in one factory's clients first. But it's going to be around 200 customers to start with.

Stephen: Great. OK.

Can: I'll send you a list of core features. So to sum up — I'll test the email feature one last time just to make sure it's working. And then I'll reach out to some people to do some demos with them in Quebec.

Stephen: Yeah.

Can: I'll do some videos. Probably in English. Just to tease what's coming. I'll create an Instagram for this. Probably the logo isn't going to be ready in time. But maybe we'll have a simple logo so we can at least launch the Instagram.

Stephen: But don't worry about branding. Just go to Nano Banana or I can even do it. I can get Friday to use the API and start hammering out logos, because she knows the codebase so well. She knows our ICP, she knows everything from these discussions. She can just hit the Nano Banana API and I can get a folder of logos. I'll get her to iterate them. You tell me if they're rubbish or not. She knows the colour palette already. She knows all this stuff. She can do it.

Can: Would you like to? I tried doing that. We got a few logos. I think I used them for the startup application. Some of them. I wasn't in love with any of them. I'm not trying to be picky. It's like maybe I have something in my mind. I feel like a logo is important, especially in Quebec. These restaurants are heavy on branding. That's how we sold the wine. If we are launching, I want to launch with a logo that I'm very happy with.

Stephen: Yeah, well we can have a placeholder, right? You don't have to have any official launch. You can always iterate on it. It's better to launch rough than not launch at all — than wait for perfect.

Can: Hundred percent. Yeah, hundred percent. Cool. All right Stephen, thank you very much for this.

Stephen: No problem. And I'll send you the summary. Have a good weekend.

Can: All right, thank you.