Bespoke ERP, without the development project.
Software shaped around how your business actually works, for firms in the UK and Spain. AI now writes the code. I bring twenty years of ERP judgment to decide what it should do, a platform engineered to build it reliably, and I run it once it’s live.
From conversation to system
-
You say
“If a pallet turns up damaged, we claim it back from the supplier. We never just write it off.”
-
We agree
Damage found at goods-in raises a claim against the supplier, linked to the purchase order. The damaged stock leaves inventory at cost, and the claim is set against their next invoice.
-
It’s live
Journal entry GRN-2291 Account Debit Credit Supplier claims 420.00 Stock 420.00 420.00 420.00 Balanced, and in your books.
AI writes the code. The hard part is knowing what to build.
Software shaped around your business beats bending your business to fit a package. It used to mean years of developers writing code by hand, at a price few smaller firms could justify. AI has removed most of that cost. What it hasn’t removed is deciding what to build, and running it properly once it exists. That’s what I do.
Working out what to build
An ERP system encodes decisions about your business: how stock is valued, when revenue is recognised, who approves what, and what happens when an order goes wrong. Get those right and the software follows. Get them wrong and no amount of code will save it.
We meet regularly. You walk me through how the business really works, including the exceptions and workarounds that never make it into a requirements document, and we decide how the system should handle them. Often that means questioning how something is done today, not just copying it.
Sessions run on a meeting tool I built for this. An AI assistant follows the conversation, already knows your system as it stands, and keeps a shared list of what we’ve agreed. Nothing is marked as agreed until I confirm it. Afterwards, our decisions become a precise written specification in your business’s own language, and over time those specifications become a complete record of why your system works the way it does.
A platform built for ERP, not for everyone
Your system is generated on a platform I’ve built and maintain: a deliberately chosen technology stack, and an automated pipeline that turns agreed specifications into working, deployed software.
The pipeline turns each specification into a structured build plan and holds the AI to automated quality checks it can’t talk its way past. Independent AI reviewers then audit every piece of work without seeing how it was produced. Nothing reaches your system until it passes, and as AI improves, so does the pipeline.
- SpecificationAgreed in our session
- Build planBroken into steps
- AI buildCode generated
- Quality gatesAutomated checks
- Independent reviewAI reviewers audit the work
- LiveIn your system
| Technology | Why it’s there |
|---|---|
| Go | Fast and dependable. It compiles into a single self-contained program, so there’s little to go wrong. |
| SQLite | Among the most thoroughly tested software ever written. Your whole database is one file you own, supported until at least 2050. |
| React and TypeScript | The most widely used interface technology, so developers who can work on it are easy to find. |
| Open source | Throughout. No vendor lock-in, no subscriptions underneath, and nothing stopping another developer taking over. |
Run for you, properly
Once your system is live, keeping it running is my job. You don’t need an IT department or a hosting contract of your own.
- Hosting
- Professional servers in European data centres, with encrypted connections as standard.
- Backups
- Multiple independent copies, kept over different periods and in separate locations, including off-site. No single failure can take your records with it.
- Monitoring
- The system is watched continuously, errors are logged, and problems raise alerts. I usually know about an issue before you do.
- Updates
- Features, fixes and security updates go live through the same pipeline, without disruptive upgrade projects.
- Incidents
- Handled by the person who designed your system and knows your business, not a support queue reading from a script.
Where the effort goes.
Traditional custom ERP spends most of its budget on people writing code, and leaves you depending on whoever wrote it. This puts the effort where it makes a difference.
| Traditional custom ERP | With me | |
|---|---|---|
| Who writes the code | Developers, by hand | AI, on a pipeline engineered for ERP |
| Where the budget goes | Mostly into writing code | Deciding what to build, and running it well |
| When changes arrive | At the end of a project phase | Shortly after we agree them |
| Who owns the code | Often the developer or vendor | You, from day one |
| Where your data lives | Inside someone else’s platform | In one database file that belongs to you |
Who you’d be working with.

“Most of what goes wrong in ERP is decided long before anyone writes code.”
I’m Jonathan Pincas. I’ve spent twenty years designing ERP systems across accounting, inventory, ordering, logistics and compliance, for businesses in the UK and Spain. Before this practice I built Pakk, an all-in-one commerce and operations platform with double-entry accounting at its core.
UK and Spanish requirements, including Making Tax Digital and VeriFactu, are part of the design from the start rather than added later.
Tell me what isn’t working.
A short note is the best way to start: what’s broken, what you’ve tried, and what good would look like. English or Spanish.
Prefer to talk it through? Book a call, or email jon@pincas.eu.
Or start with the two-page overview PDF
Based in León, Spain. Working with clients across the UK and Spain.