Startup's Viral 'Decision' AI Model Jev Draws Swift Copycats From OpenAI and Amazon

A small startup called TypeSafe AI has managed to do something few companies can: spark a genuine viral moment among developers with a brand new kind of AI model, only to watch some of the biggest names in the industry rush out copycats within weeks.

The model, called Jev, was released on September 15 by TypeSafe AI, a company founded by Diogo Almeida, a former OpenAI researcher. Rather than generating text word by word the way conventional large language models do, Jev is built specifically to make fast decisions, returning a structured output such as a choice between options, a numerical score, or a yes or no probability in a single parallel pass rather than through token by token generation.

A Speed Advantage That Caught Developers' Attention

That design gives Jev a dramatic speed advantage for the narrow task it is built for, responding in roughly 70 to 500 milliseconds and running somewhere between 40 and 200 times faster than conventional approaches for decision style tasks. The model spread quickly after becoming available through Vercel's AI Gateway, with nearly 13 percent of Vercel's paid development teams trying it within a single day, a faster initial adoption rate than any previous model launched on the platform. TypeSafe AI has said roughly a quarter of Fortune 500 companies have since put Jev to use in some form.

Big Tech Moves Fast to Catch Up

The viral reception did not go unnoticed by larger competitors. On October 6, just three weeks after Jev's debut, OpenAI launched its own competing product, a tool called Decisions API aimed at the same kind of structured, fast turnaround decision making. OpenAI was not alone. Within sixteen days of Jev's release, Amazon, Cloudflare, and Databricks had each shipped comparable products of their own, while independent developers published dozens of open source alternatives attempting to replicate Jev's approach.

The rapid response from some of the industry's largest players illustrates how quickly a promising idea can be absorbed once it proves itself with real usage, leaving a small startup like TypeSafe AI to defend the lead it built with Jev against far better resourced rivals racing to match it.