AI Solutions: AI application
Valoov
A company valuation in minutes, with the model doing the talking and the code doing the arithmetic
The problem
A valuation is a number someone may act on. A confidently wrong one is worse than no number at all. That makes it the wrong job to hand to a language model end to end, and the right job to split.
What we built
- Google Cloud throughout. Vertex AI for the model layer, Dialogflow CX for the guided intake conversation, Python for the valuation engine.
- A structured interview collects what the method needs, then returns a valuation immediately rather than a callback.
- Serves the French and Spanish markets, so the conversation has to cope with how founders describe a business in more than one language and accounting convention.
Where the automation sits, and where it does not
Conversation, intake and explanation are model work: they have to cope with how a founder actually describes a business. Method selection, the arithmetic and the audit trail stay in deterministic code. The model asks and explains; the code computes and records.
What we would not claim
No accuracy benchmark, no user numbers and no client-attributed revenue. A valuation platform invites exactly those claims and we have not measured any of them.
Client engagement. Platform read September 2026.
Your turn
We set the baseline before the work starts.
That is the part that makes a number mean something a year later.