Skip to content

Ex-OpenAI Researcher Launches TypeSafe’s Jev, Claiming Zero-Hallucination AI for Preset Decisions

Sep 16, 2026
The Rundown AI
Article image for Ex-OpenAI Researcher Launches TypeSafe’s Jev, Claiming Zero-Hallucination AI for Preset Decisions

Summary

Ex-OpenAI researcher Diogo Almeida unveils TypeSafe’s Jev, an AI for pre-defined software decisions that claims zero hallucinations by selecting only from preset options and returning confidence scores.

Key Points

  • Ex-OpenAI researcher Diogo Almeida’s TypeSafe emerges from stealth with Jev, an AI system for preset software decisions that attaches confidence scores and claims zero hallucinations.
  • TypeSafe prices Jev at $42 per billion input tokens with free output, estimating it is 238 times cheaper than Claude Fable 5.1, and says responses arrive in 70 to 500 milliseconds.
  • Jev chooses only among options defined in advance and targets tasks such as sorting requests, scoring records, and screening other AI outputs for jailbreaks.

Tags

Read Original Article