Your industry is not on our list. Here is how we learn it.

Four problems that come up whatever the sector. None of them needs industry knowledge to recognise, and all of them need it to fix well.

1. 01Every team reports a different number

   Finance, sales and ops each pull their own extract, so the meeting starts by arguing about the data.

2. 02Decisions wait on last night's batch

   The warehouse refreshes overnight, so Monday's call runs on Friday's numbers.

3. 03The real work still runs on spreadsheets

   The system of record says one thing, the spreadsheet and the WhatsApp group say another.

4. 04The legacy system blocks every change

   Ten years of custom logic nobody wants to touch, and every new requirement has to get past it.

The industry knowledge has to come from somewhere. In the first weeks it comes from you, in this order.

1. 01Two calls, and the three decisions that matter

   The Week 1 audit maps where your data sits and how it moves, the top three decisions the business needs it to support, and where the friction sits today.

2. 02Time with the people who will use it

   Before anything is drawn, time with the person who will use it: what they open, what they ignore, and what they pick up the phone about instead.

3. 03Your words, agreed once

   The terms your business uses, and what each number means, written down and agreed once, where people can find them.

4. 04Your own examples, graded

   Where AI is involved, a graded set of your own examples. Public leaderboards say nothing about your abbreviations, your scanned documents or your edge cases.

5. 05Something real by week three

   Software a real user can open, early enough for your team to disagree with us while disagreeing is still cheap.

Three pieces of our published work where the domain knowledge came from the client, and how it got into the build.

1. ManufacturingUloborus, built with a working factory

   Every module was built against how Rajaram Consumer Care in Sangli actually works, then argued about, then changed.

   See Uloborus ↗
2. TradingStockJarvis, where the AI had to learn trading logic

   Domain-specific prompt frameworks, designed and refined so Claude correctly interprets complex trading logic, entry and exit rules, and risk parameters.

   Read the StockJarvis story ↗
3. Design studioDesign Asylum, a different client every project

   Ten-plus projects since 2025, for clients across the sectors the studio serves, each with its own stack, integrations and data.

   Read the Design Asylum story ↗

The same three phases in every industry. What changes is what we build in them.

1. Week 1, fixed feeAudit

   Two calls and a written one-pager. Yours to keep whether or not you continue.

2. Weeks 2 to 6Build

   Something real in front of real users by week three. Median six weeks to the first production-grade artefact.

3. Week 7 onwardOperate

   Quarterly reviews and on-call governance. The people who built it pick up the phone.

• The same customer under four names breaks every number downstreamA problem we have written about for three industries.
• You cannot choose a model without a graded set of your own examplesA hundred graded examples of your own say more than any leaderboard.
• You cannot automate a process nobody has written downThe flowchart you were handed describes the good day.