Decision-Making AI Model Jev Goes Viral in Silicon Valley

The AI model Jev, launched by startup TypeSafe AI, has become a hot topic in Silicon Valley just three weeks after its release. Unlike chatbots, it does not write text. Instead, it uses machine learning to sort inputs into preset results, such as a yes-or-no answer or a score. The company calls this reinforcement learning for calibrated decisions.

Jev was built by Diogo Almeida, who worked on ChatGPT at OpenAI. His company has raised $40 million. He says about 25% of Fortune Global 500 firms are using Jev, which now handles one trillion tokens a day. The company is also in talks for new funding at a valuation above $10 billion.

The buzz has inspired many similar products. OpenAI launched a Decisions API powered by its Luna model, Databricks rolled out ai_decide, and Cloudflare released open-source Clef models. Amazon offers a similar tool too. A new model category is forming: generative models handle complex reasoning, while decision models handle sorting and approvals.