Built after watching a year of records get reconciled in one week.
Profinlex started from a scene most finance teams recognise: a conference room in late March, printouts across the table, a team of qualified people tying one list to another because the system of record could not answer a question it should never have needed to be asked.
- What we build
- An AI ERP for compliance-heavy businesses
- Where we focus
- Reporting rigor, not operational reporting
- How we sell
- Compliance mapping first, quote after
The scramble is not caused by ignorance. It is caused by architecture.
Every team in that room knew the standards. They knew which input credit was ineligible, which related-party transaction needed an approval it never got, which lease had been expensed when it should have been recognised. What they did not have was a system that applied any of it on the day the entry was passed.
So the knowledge sat in people, and the enforcement sat in a deadline. Under that arrangement the work always collapses into the same week, and the errors that surface are months old - already inside reports that were circulated, already inside returns that were filed.
Existing software did not help, because it was built for a different question. Operational ERPs optimise throughput: how fast an order becomes an invoice. Accounting software optimises recording. Neither one tests whether a treatment holds against Ind AS, whether a credit is eligible under s.16, or whether an approval preceded the transaction it was meant to authorise.
Profinlex was built around that missing test. The operational modules exist because verification needs the transaction, its document and its authority at the same moment - and you cannot get that from an integration running overnight. The compliance layer is not a report on top of the data. It is a condition the data has to satisfy before it settles.
It is the third ERP we have built. The earlier two were shaped around trade operations and around manufacturing flow, and both taught us the same thing: the businesses under the heaviest regulatory load were being handed a reporting module and told it was compliance. That is the gap this product is for.
Three statements we design against.
Each one has cost us a feature request. That is broadly how we know they are load-bearing.
Audit-ready always, not just in March
A compliance position that is only true at year end was never a position. It was a reconstruction. Everything in the product is arranged so the readiness figure means something on an ordinary Tuesday in August.
Every number verified, not just recorded
Recording a transaction accurately is table stakes; any accounting system does it. The question that matters is whether the treatment holds against the standard that governs it - and that has to be answered when the entry is made, not when it is questioned.
Compliance should not depend on one person remembering
In most finance teams there is one person who knows the calendar. When they are on leave, or they resign, the risk is immediate and nobody can quantify it. Obligations belong to a register with named owners, not to institutional memory.
Four things we have been asked for and declined.
A compliance product earns its position by what it refuses as much as by what it ships.
We will not auto-file without approval
A return leaving your system is a legal act. Profinlex prepares it, checks it and hands it to a named person. The last click belongs to someone who can answer for it.
We will not present an opinion as a certainty
Where a treatment genuinely needs judgment - a provision estimate, an arm's-length position, a contested classification - the line is held for a person. We would rather flag more than we resolve than resolve something we should not.
We will not build a compliance score that flatters
The readiness figure counts only what has actually passed. Exceptions are not weighted down to keep the number looking healthy, and there is no setting that hides them.
We will not claim a certification we do not hold
Our security posture is published as three lists: in place, in progress, and not claimed. The third list stays on the page even when it is commercially inconvenient.
Do not take our word for the product. Run it against a year you already closed.
We publish no testimonial wall and no logo strip, because neither tells you whether the verification rules are any good. The validation exercise does: one historical year, the full rule set, and a written list of what it found. If the findings are trivial, you probably do not need us yet.
If the close is the worst week of your quarter, that is the conversation we want.
We are more useful discussing a specific failure than describing a product. Bring the return that got filed late, the credit that was reversed, or the approval that came after the payment.
- A working session on your own close, not a slide deck
- Straight answers on what the product does not do
- Findings from one historical year before any commitment