Weflayr

Weflayr is a Paris start-up, based at Station F, that builds a routing and evaluation layer for companies that run AI models in their products. Its view is that existing AI gateways switch between model providers without knowing how each model performs on a particular company's workload, which keeps most companies tied to a few large models. It announced a $2.5 million pre-seed round in October 2026.

What the product does

Weflayr works in four steps. A customer benchmarks candidate models against its real production calls, scored on its own quality and cost objectives. The customer then approves the models it trusts and sets a routing strategy. For each incoming request, Weflayr finds similar calls it has already benchmarked and sends the request to the best-performing approved model within those limits. It also ties every inference decision to its business context, so costs and margins can be seen per customer, feature and workflow. In its first deployment, Weflayr says, inference costs fell by more than 55% at equivalent quality.

Who it serves

Software companies that build AI features and spend heavily on inference, and that want to use more models without risking output quality. The product was in private onboarding of first customers in October 2026, with wider availability planned and a waitlist open.

Key facts

Founders: Romain Mazeries, formerly chief executive of Mangopay, and Noé Achache, formerly head of AI engineering at Theodo Data & AI Base: Station F, Paris, France Funding: $2.5 million pre-seed announced 22 October 2026, led by 13books Capital with Founders Future, Kima Ventures, Plug and Play and Station F Ventures Status: first customers onboarding; waitlist for wider access