We are building the accounting system that can explain itself.

Alfrd reads your ledger, bank and inbox, works out how the business actually operates, and produces accounting work with the evidence still attached. A qualified person approves it before it counts.

Where Alfrd started

Two people, two routes in, the same broken thing underneath.

Founder note: Darshan

DarshanCTO

Background

Trained in accounting and finance first, then in software. The unusual order is most of why this is built the way it is.
The first real financial data I tried to model came out of a shipping business running across several countries.

What I found

It could not answer a single question I wanted to ask it. Nobody had been careless with it. It had simply never been built to be asked anything.
Every question that mattered needed a fact that was not in there.
I assumed I had been handed a bad example. I had not.

Why I am here

None of that was a failure of the software. It was the ceiling of what software could do, and that ceiling has just moved.

Founder note: Sammi

SammiCEO

Background

Raised money and started my first company at eighteen, and never had a clear view of its finances.
I cared about it. Looking properly meant hours inside tools built for accountants, and those hours were never there.

What I found

The company did not survive, and not knowing where the money was is part of why.
Years later I was building with AI agents and walked into the same wall from the other side.
The agents were fine. What I was handing them was scattered across systems, inconsistent between them, and months out of date.

Why I am here

The bottleneck was never how good the model is. It was the ground underneath it, and that ground had already cost me one company.

Neither of us set out to build this. We got to the same answer from opposite ends, one from having to produce the numbers and one from having to rely on them, and when we compared notes it turned out to be the same problem written down twice.

What we have learned from the people who do this work

We have been putting this in front of practising accountants since the first week it was usable, and those sessions have changed the product more than anything we worked out on our own.

One has used Xero daily since 2009 and was a partner in a sixty-person firm before starting his own implementation practice. He told us the gap we had picked is real and close to unserved, and that in sixteen years nobody had ever asked him for a second ledger.

Another is an FCCA, a fractional CFO and a non-executive director, and was an investment manager at a venture fund before that. She was harder on us and more useful for it. She had already automated the straightforward half for a charity she advises, using general-purpose tools, and stopped deliberately at the point where sensitive data began. Her argument was that the half a capable person can assemble alone is not the half worth paying for. She was right, and it reordered what we are building.

We would rather hear that in a call than find it out at a renewal.

Where we think this goes

The part of this job that needs a person is the judgement: whether a contract changes how revenue lands, or whether a cost belongs in this period rather than the last one. Nearly everything you do to get to that point does not need a person, and that is the part we are trying to delete.

We are building for EMEA first. Most of this field is built for US accounting and treats everywhere else as something to localise later.

We started with journal entries because every number in a forecast or a set of accounts is a sum of them. What sits underneath is a model of how one business actually works, and most of that is still ahead of us.

See open roles