
Weflayr raises $2.5M to route AI requests using production benchmarks
The AMW Read
Weflayr adds an early-stage inference-routing entrant, but one reported deployment supports only a limited update within the infrastructure segment.
Weflayr raises $2.5M to route AI requests using production benchmarks
Paris-based Weflayr closed a $2.5 million pre-seed round led by 13books Capital. Participants include Founders Future, Kima Ventures, Plug and Play, Station F Ventures, General Catalyst scout Roxanne Varza and angel investors. Founders Romain Mazeries and Noe Achache previously led Mangopay and AI engineering at Theodo Data & AI, respectively. The funding will expand engineering and AI research teams and support growth across Europe and the U.S.
Weflayr targets the model-selection layer of AI infrastructure, where teams must balance quality, speed and cost across competing models. Its system benchmarks models against customers' production data, lets teams approve models and define routing rules, then matches requests to similar benchmarked calls. It also connects inference choices to costs and margins across customers and features. The market question is whether workload-specific evaluation can make an independent routing layer valuable beyond basic gateway functions. Weflayr reports that its first deployment reduced inference costs by more than 55% at equivalent quality. The report does not provide the evaluation method or workload details needed to assess that result.
For builders, the concrete implication is to evaluate routing against their own production tasks and quality requirements before adopting reported savings as a target. Weflayr's customer- and feature-level margin tracking could help teams connect model selection to product economics. For investors, the initial deployment offers an early signal; the article provides no evidence that the savings repeat across customers or workloads.