Sign InOpen Brain
VercelEngineering PostOfficial Source

How Searchable ships customer-requested features in 30 minutes on Vercel

Searchable says Vercel’s AI SDK and Gateway let it swap models without SDK or key changes, helping the team ship some customer requests in 30 minutes and move 2–5× faster.

Vercel · Jul 21, 2026
Open Source Open MarkdownOpen JSON
Source Summary

Searchable has processed **100+ billion tokens** and says some customer-requested features ship in **as little as 30 minutes**. Its team reports development moving **2–5× faster** after building on Vercel’s AI SDK and AI Gateway.

Practical Implication

For an agent product that tests several providers, a unified SDK and gateway can remove model-specific integration and API-key rotation from each experiment. That shortens the path from evaluating a model to putting it behind a feature.

Agent-Ready Context
Searchable has processed **100+ billion tokens** and says some customer-requested features ship in **as little as 30 minutes**. Its team reports development moving **2–5× faster** after building on Vercel’s AI SDK and AI Gateway.

For an agent product that tests several providers, a unified SDK and gateway can remove model-specific integration and API-key rotation from each experiment. That shortens the path from evaluating a model to putting it behind a feature.

These are Searchable’s reported outcomes, not a controlled comparison. The material does not separate the gateway’s contribution from team process, application architecture, or other changes.
Context Map
infracoding#gateways#agent-sdks#adoption
Uncertainty
These are Searchable’s reported outcomes, not a controlled comparison. The material does not separate the gateway’s contribution from team process, application architecture, or other changes.