Loading page…
A consultant rebuilt a food wholesaler's dispatch and reordering with Claude Code, refused to invoice for it, and took a net 1% of the extra turnover it produced
his own consulting hour is priced at about €400 · his firm's website took 78 hours of his time, was costed internally at €3,000, and runs on €9/month hosting, against agency quotes of €6,000–€7,000 to build the equivalent · €2,500 asking price for a six-hour Notion CRM setup (quoted, not yet sold) · first sales roughly three months after he started learning
"I didn't ask for any money. I said: give me a percentage of the profit difference you're going to make. Net 1%." The client's turnover went from about €2 million to about €2.8 million in a year.
Fewer bars = easier, cheaper, or faster for an AI-assisted solo builder. Editorial judgments based on the case details.
Aksel is a consultant who sells his hours at around €400 and who, by his own account, did not understand what any of this automation material was for until he started using it on his own work. The interview is conducted by the person who trained him, inside that trainer's community, which is worth knowing when weighing the numbers. He starts with his own website. He built every character of it with Claude Code — a separate agent for copywriting, AI generation for every image, no human-written text or photography anywhere — spent 78 hours on it, and pays €9 a month to host it. He costed the work at €3,000 against his hourly rate. Then he took it to the web agencies he knows, told them a site like this had been built for €10,000, and watched one offer to do it for €6,000 and another for €7,000. His own competitors, firms older than his, are running plain WordPress pages. The second piece is the one that matters commercially. He adapted an automation his trainer had originally sold to hotel operators and pointed it at a food wholesaler's dispatch problem: collect the orders, assign delivery regions, pick the right truck size, sequence the stops. It replaced roughly three hours of planning a day. He did not invoice for it. He asked for a net 1% of the increase in the client's turnover. The client was doing about €2 million around the same period last year; it is now close to €3 million, which he immediately rounds down himself to about €2.8 million.
He sold to the businesses already around him. The buyers named are local food wholesalers serving an ethnic grocery segment, a grocery retailer in that same supply chain, and a firm making doors and windows — none of them found through marketing.
The quick-reference card — acquisition channels, replication playbook, risks & traps, and our verdict — is members-only.
Join the Unicorn Club — $5/moCancel anytime, in one clickPaste into Claude Code / Codex and get a working version of this product end-to-end.
Two production-grade prompts per idea — a full build spec and an SEO growth plan — plus 10 tailored build specs a week. Members only.
Join the Unicorn Club — $5/moCancel anytime, in one clickEvery claim traces back to a listed source — open any of them and check the number yourself.
Data credibility: Founder-reported inside a promotional context. Every figure comes from Aksel speaking in an interview conducted by the trainer whose paid community he belongs to, published on that trainer's channel, and the host explicitly frames the outcome as validation of his own teaching — both parties benefit from the story being impressive. Nothing is documented: no contract, no invoice, no dashboard, no client name. The client's growth from about €2M to about €2.8M in turnover is his relay of the client's own number, and he volunteers that he is rounding. The causal link between the automation and that growth is asserted by the host and accepted in the conversation, not demonstrated — what is actually evidenced is the removal of roughly three hours of daily planning work. The 1% is described as a net percentage of the turnover difference with no payment schedule or audit right mentioned; one percent of an €800,000 increase is roughly €8,000, which is our arithmetic on his figures, not a number he states. His own annual or monthly income is never disclosed. The €2,500 Notion CRM price is a quote for a meeting scheduled for the following week, not collected revenue. The €3,000 website figure is what he costed his own firm's site at using his hourly rate — it was never sold to anyone; the €6,000 and €7,000 figures are counter-quotes from agency friends after he told them the site had cost €10,000. The automatic captions drop leading digits from several euro amounts (rendering figures as '.000 euro'), so comparable-system prices mentioned in the interview are described qualitatively rather than quoted. The interview is in Turkish and all pricing is in euros; he never states which country he lives or works in, so none is claimed here.
It took about three months from starting to learn to making his first sales.
Two loops, and the second is the interesting one. The first is ordinary: fixing one process exposes the next.
His quiet stretch was three months long and it was a learning period, not a sales drought. He describes starting with ChatGPT, moving through Antigravity, and then Claude Code arriving mid-journey and changing the approach again — three months of tool churn before he made a single sale.
Four distinct deliverables, built by one person. The dispatch system is the flagship. Four stages: collect the orders, decide which regions each customer falls into, pick the truck size against the volume, and sequence the stops from first customer to last with an expected finish time. The output goes to the driver's phone as a Telegram conversation — the customer list, with addresses that open straight into Google Maps. The driver replies as he goes: delivered, this many pallets dropped, this many pallets collected, this payment taken. In the evening the manager receives a report of everything that driver delivered, the pallets returned and the collections made. The reordering system sits upstream in the same supply chain. He requested the retail software vendor's API — new this year, unavailable last year — and connected it so that when shelf stock depletes past a threshold, orders accumulate into a bundle in a spreadsheet that is emailed to the relevant wholesaler every two weeks. The third is the Notion CRM: order tracking, order management and payment follow-up for a joinery business that was running all of it in a spreadsheet, plus chat notifications when something is outstanding. The fourth is his own marketing site, built entirely with Claude Code using a dedicated copywriting agent and AI-generated imagery for every section, with the flow modelled on large consulting firms' websites. Our read: none of these are products. They are four instances of the same service — find a chain of repetitive decisions, encode it, and deliver the output where the worker already is, which in every case here is a chat app rather than a dashboard nobody opens.
Three pricing modes, deliberately different. Hourly, at about €400, for consulting — the anchor that makes everything else look cheap. Fixed-price installation, at €2,500 for a six-hour Notion CRM build, quoted for a meeting the following week and therefore not yet collected. And revenue share: a net 1% of the increase in a client's turnover, which is how the wholesale dispatch system was sold. The revenue share deserves the attention. He says he asked for nothing up front, only a percentage of the difference in turnover the client would make, and the client agreed. The client's business went from about €2 million at the same point last year to close to €3 million, which he himself immediately rounds to €2.8 million. One percent of an €800,000 increase is roughly €8,000 — that is arithmetic on his two stated figures, not a number he gives, and the payment schedule is never described. He also gestures at what comparable turnkey systems cost, likening it to buying a terminal-style system sold in the tens of thousands of euros and saying he could sell a maintained version at half that. The captions drop leading digits from those particular figures, so we are not quoting them as prices. Our read: the revenue share is the right instrument here for a reason that generalises badly. It works because dispatch and reordering sit directly on top of the number being measured, so the client accepts the causal story. Try the same deal on a marketing automation or an internal reporting tool and you will spend the relationship arguing about attribution.
The running cost of the flagship website is €9 a month for hosting — he mentions moving up a hosting tier because the site is heavy, and the upgraded figure is garbled in the captions. The build cost was 78 hours of his own time and nothing else: a separate Claude Code agent for each job, one dedicated to copywriting, AI image generation for every tab and heading, and no supplied copy or photography of any kind. The single manual intervention was language quality — the output read like machine translation where it needed to read as native, and fixing that took him about twenty minutes of re-prompting. The automations run on equally cheap infrastructure: a Telegram bot as the driver interface, Google Maps links for navigation, a spreadsheet as the order bundle, email as the delivery mechanism to wholesalers, Notion as the CRM, and a third-party retail software API he obtained by asking the vendor for it. Our read: margins here are effectively the whole invoice, because the cost of goods is the operator's hours and the tooling is a rounding error. That is also the strategic vulnerability. When the delivery cost of a system collapses to six hours and €9 a month, the only thing holding the price up is the buyer's ignorance of that fact — which is exactly the arbitrage he demonstrates by getting €6,000 and €7,000 quotes for a site he built himself. That window closes as buyers learn, and the operators who survive it will be the ones selling measured business outcomes rather than builds.
Copy the compensation structure, but only where it fits. A net percentage of the increase works when your system sits on the metric — dispatch capacity, order volume, collections — and when the client's growth is plausibly attributable to it. Ask for a slice of the delta rather than a fee, and you convert a one-off invoice into income that grows without further selling. Copy the vendor API move. The retailer's software company had an API that had only existed since this year, and he got it by asking. Most small-business software has an integration surface nobody in the building knows about; one email to the vendor is the cheapest unlock in this entire playbook. Copy the delivery surface. The driver never opens an app — the work arrives as Telegram messages and he answers in the same thread. Automation that requires a field worker to log into a dashboard does not get used. Change the trust arrangement. A revenue share means seeing the client's books, and he notes that not everyone grants that access; no contract, audit right or payment schedule is described anywhere in the interview. Put the measurement method, the baseline and the payment cadence in writing before you build. Change the client concentration. One rev-share client and a handful of small setups is a practice with a single point of failure. The stable version pairs two or three fixed-fee installations a quarter with one outcome-based deal, so the annuity is upside rather than rent.
Attribution is the biggest one. The client's turnover moved from about €2 million to about €2.8 million, and the interview treats the automation as the cause. There is no counterfactual, no breakdown of how much of that growth came from new customers, pricing, market conditions or the owner's own sales work, and no third party has looked at the books. The automation demonstrably removed about three hours of daily planning; converting that into €800,000 of additional turnover is a story both parties in this conversation have an interest in telling. The deal structure carries its own risk. A handshake percentage of a growing number invites renegotiation precisely when it starts paying well, and nothing in the interview describes a contract, an audit right, a baseline definition or a payment schedule. The client also controls the measurement, which is an uncomfortable position for the person being paid from it. Evidence context matters here. This is an interview conducted by the trainer whose course and community the guest is a member of, published on that trainer's channel; the host repeatedly frames the outcome as validation of his teaching. The guest's own total income is never stated — not for the year, not for the month — and the €2,500 CRM price is a quote for a meeting that has not happened. The automatic captions drop leading digits from several euro figures, which is why some of them are not repeated here as prices. Operationally, the reordering system depends on a third-party retail software API that only appeared this year and could change or be withdrawn, and the driver workflow depends on a consumer chat app. Both are outside his control. Finally, the arbitrage itself is temporary. Charging against a €6,000–€7,000 agency benchmark for work that costs six hours and €9 a month is profitable exactly as long as buyers do not know what it costs. That is a window, not a moat.
The four founder playbooks — first customers, first dollar, the flywheel, the silent stretch — are in the Unicorn Club, on this idea and on all 1,435 others. The story, the numbers, the sources and every later section stay open to everyone.
Join the Unicorn Club — $5/moCancel anytime, in one click