Our thesis
We don't sell software. We sell the work it does.
For thirty years, enterprise software sold you a tool and left the work to your team. You bought seats, and then you hired people to operate what you bought.
BLAYER builds the opposite. AI employees that do the operational work inside your company — under rules you set, with a person accountable for every decision that matters, and a record of everything they did.
We call it software in a service. The software is how it reaches you. The finished work is what you're buying.
Why software you operate is ending
The tool was never the point.
Nobody wanted a reconciliation screen. They wanted the reconciliation done. Software sold the screen because that was all it could sell.
Capability stopped being the scarce thing.
Models can already read an invoice, match a payment, and draft an explanation. What stops a company handing that work over isn't capability. It's that nobody can prove what the system did, why it did it, or who is accountable when it is wrong.
So we build the part that makes it safe to hand over.
Rules the company sets, enforced in code. Evidence behind every claim. A person in charge of anything consequential. A permanent record of all of it. That is the product. The AI is the commodity.
How it works
- 01
It runs where you do.
Installed in your environment. Your records, documents and communications stay inside your walls. We don't hold them and can't retrieve them.
- 02
It learns how your company works.
Your chart of accounts, your close calendar, your approval limits, your policies, your vendors, your history — held as a private model of your operations that you can read and edit.
- 03
AI employees do the work.
Narrow, named jobs with defined scope: reconcile this, prepare that, chase this document, draft that explanation.
- 04
Policy decides what they're allowed to do.
Not the AI's judgment — a rule you set, enforced in code. It recommends; policy decides; a person approves anything material. Money never moves unattended, at any setting.
- 05
Everything is recorded.
What data was used, which rule applied, what was checked, who approved it, what changed. Click any figure and see where it came from.
What's real today
BrainB, our first product, runs finance. Month-end close and reconciliation, for mid-sized companies, deployed on their own machines. It's in development with no customers yet — we're looking for the first few finance teams willing to tell us what we've got wrong.
What's built
- A deterministic policy engine
- An evidence layer behind every finding
- An append-only audit trail
- Nine finance agents
- A double-entry general ledger
- Multi-entity consolidation
- ASC 606 revenue schedules
- Canadian payroll
- 821 automated tests
What isn't, yet
- Live integrations to your systems
- A security certification
- A single paying customer
- An incorporated entity with insurance
We'd rather say that than imply otherwise.
Where it goes
Finance is first because it is the hardest: accurate to the cent, heavily controlled, full of sensitive data, and it touches every other department. If the architecture survives finance, it survives anywhere.
The same environment extends to procurement, legal, HR and operations — the same rules, the same evidence, the same record. A company doesn't buy five disconnected AI tools. It runs one governed AI workforce.
Safety
How BrainB stays safe
- 01No money moves unattended.
- 02No payment is initiated, changed, or released without an authorised person approving it at the moment of execution.
- 03No bank account or vendor payment detail changes without high-risk verification.
- 04No financial statements are signed off by software.
- 05Material or non-routine journal entries require Controller approval.
- 06External communications require authorisation.
- 07A single correction never silently rewrites a policy.
- 08It gets only the data its specific job requires.
- 09Every recommendation links to its source.
- 10Every consequential action is recorded in full — what data was used, what was decided, who approved it, what happened.
I'm Mitansh. I started BLAYER on one bet: the era of buying software and then hiring people to run it is ending. What replaces it is work delivered as a finished outcome — under rules the customer sets, with a person accountable for every decision that matters, and a record of all of it.
I'm early. No customers yet, no certification, no incorporated entity with insurance. I'd rather tell you that than imply otherwise — and I'd rather a finance team tell me where I've got this wrong than keep guessing for another six months.
Mitansh · Vancouver, Canada