Why we build AI that shows its work

AI tools have become very good at sounding confident. In research and in finance, that is not enough. A thesis chapter needs citations that point to real sources. A revenue figure needs to trace back to a contract clause. An investment plan needs rules that hold up against the evidence. When an answer cannot be checked, it cannot be trusted, and someone still has to redo the work.

That is the problem AI Central Lab was founded to work on. We are a software company founded in Québec in 2026, and everything we build follows three rules.

AI proposes, people decide. Our products use AI for the slow parts of the work: reading documents, finding literature, extracting figures and suggesting mappings. Every suggestion waits for a person to accept, edit or reject it, and that decision is recorded.

Every output traces to its source. A figure links to the page it came from. A sentence links to the reference it cites. A revenue number links back to the contract. If the evidence is not there, the tool says so instead of filling the gap.

Deterministic where the numbers matter. Calculations run on tested rules, not on a language model, so the same input gives the same answer every time.

Where we are today

Our portfolio covers research and academia, accounting and audit, personal investing and financial data. ThesoraX, our research workspace for theses and papers, is in early access and patent pending in Canada. RevTrace AI, for IFRS 15 and ASC 606 revenue recognition, is preparing for a pilot programme. DoSnip is in early access for Excel on Windows. WealthMaster, a research and planning tool for self-directed Canadian investors, is in pre-release; it does not give investment advice. Our financial data conversion service is open to clients now. Our other products are still in development, and we label them that way.

What we are looking for

We are at an early stage, and we would rather say so than pretend otherwise. We are looking for pilot clients who want to test a product on real work and tell us what breaks, design partners who want to shape what we build next, and investors who share our view that useful AI has to show its work.

If that sounds like you, write to us at contact@AICentralLab.com.