Inventory management & shop
Built in-house · in daily useOrders, stock and product data in one tool, wired straight into the shop. Maintaining it means checking — not typing.
AI handlesproduct copy, categorisation, stock alerts
Case 313 · My own business
Can an entire business run on AI? Mine does — and this is what it looks like.
The starting point: anyone who recommends AI to businesses ought to be able to show what it looks like in their own. Not as a pilot, not as a demo — but in day-to-day work, with invoices, customers and deadlines riding on it. That is why my own business is Case 313: the proof on the object.
AuthorMatthias EgerCertified Manager for Applied AI Transformation (IHK)Last reviewed17 Jul 2026
To the verification recordThe solution: one business, one pipeline.
Research, concepts, code, copy, quotes — everything runs through the same pipeline. The division of labour is simple: local models where data has to stay in-house. Large models where performance counts. Confidential material never leaves my own server, and for everything else I use the strongest tool currently available.
What actually runs on AI — seven tools, built in-house:
Quotes & invoicing
Built in-house · in daily useA few bullet points become a draft quote; invoice runs and payment reminders keep a fixed rhythm. Signing stays manual.
AI handlesdrafts from bullet points, invoice runs, payment reminders
Phone assistant
Built in-house · in daily useAnswers every call, clarifies the request and summarises it — I call back prepared instead of picking up cold.
AI handlesanswering, qualifying, summarising
Inbox
Built in-house · in daily useEmails arrive pre-sorted with a suggested reply already attached. Nothing goes out before I have looked at it.
AI handlespre-sorting, reply suggestions
Development pipeline
Built in-house · in daily useThe centrepiece: code drafts, reviews and documentation in fixed slots. This pipeline built the other tools — and this website.
AI handlescode drafts, review, documentation
Research & concepts
Built in-house · in daily useMarket analyses, plans and white-paper drafts appear overnight — judging them stays manual work.
AI handlesanalyses, plans, first drafts
Translations
Built in-house · in daily useAn in-house translation tool with API connection: documents in, formatting preserved, content stays in the building.
AI handlesin-house translation, formatting preserved
And you are reading the most recent piece of evidence right now: this website is Case 313. Conceived, built and written with AI — in days, not weeks. The same pipeline I use to build projects for clients.
The result in one number
0 results leave this business without my sign-off — no text, no plan, not a line of code. The AI produces a great deal, but it publishes nothing. Every result is checked and vouched for before it reaches a client. That is what the seal stands for.
Why it works — and where the limit is.
The rule that carries the business is not a technical one but a matter of craft: AI delivers the draft, I deliver the judgement. What gets automated, what is left alone and what goes out to a client — no machine decides that. That is precisely why the rest works: if you keep the judgement, you can leave the typing to the AI without worry. It is the same rule that applies in client projects — for example with the draft quotes in the electrical trade.
“AI here is not running in a pilot project. My business runs on it.”
Transferable to your business?
This exact operating model — the human leads, the AI works — is what I set up in your business. Not as a copy of my tools, but adapted to your processes: first the routine steps that eat time every day, then the rule about who checks before anything goes out. What has to stay in-house stays in-house — and you remain the owner of everything that is created.
Want to see what this would look like at your place?
30 minutes, three starting points in writing — free and with no sales pressure.