We raised $4.2M to build the AI platform for cost intelligence

Why we started Fakto

A large enterprise signs thousands of supplier contracts. Each one is a promise written down: a price, a discount that triggers at a volume threshold, an indexation formula, a maintenance clause. Between the signature and the invoice, parts of that promise go missing. Rarely dramatically - a grid that was never applied, a rebate no one tracked, an index calculated the wrong way. At the scale of a group doing $10bn in revenue, the distance between what was negotiated and what actually gets paid runs to $50M a year.

Finance and procurement teams have lived with this. For thirty years the working assumption has been that 2 to 5% of supplier spend is simply unrecoverable - the cost of operating at scale. That assumption made sense when an audit meant a team of people and a stack of spreadsheets sampling a fraction of the contracts once a year. It does not make sense anymore.

We watched this up close in previous roles: millions in margin leaving the building every year because no one had the tools to read the company’s own contracts against its own invoices.

The reason is structural

The information that matters lives in unstructured data - free text, PDFs, pricing grids buried in Excel - and traditional software cannot read it. Large language models change that. A clause can now be extracted from a raw document, compared against the line it is supposed to govern on an invoice, and checked for deviation before the payment ever goes out.

What Fakto does

Our agents ingest a company’s entire contract estate, read every line on every incoming invoice, and flag anything that departs from what was signed. Fakto is an AI cost intelligence platform on autopilot: continuous coverage at scale, across every class of error - wrong-grid billing, indexation drift, tariff misapplications, coefficient errors, uncontracted rates.

On a first engagement covering $175M of fuel, equipment rental and MRO purchases at a major French construction group, the system ingested more than 50,000 unstructured invoices and pricing grids and surfaced 2.5% in contractual deviations within days - $2.9M in recoverable margin identified. One supplier alone was running a 14% systemic deviation rate.

The $4.2M round was led by Frst, with Global Founders Capital and Darkmode, and angels Alain Dehaze (former global CEO of Adecco), Jean-Stephane Arcis (co-founder of Talentsoft) and Dominique Vidal (Index Ventures). We chose investors who understand the French market and can move with us into the UK in the second half of 2026 and the US in early 2027. The capital goes into the product and four to five engineering hires.