TWIML · October 6, 2026 · 1h 30m

Why Jev Is Changing How We Build With AI with Diogo Almeida - #779

Diogo Almeida, co-founder and CEO of TypeSafe, discusses Jev, a model designed to bring fast, reliable intelligence into software by focusing on machine-native intelligence. He explains why text-generation models fall short for real-world automation and how Jev uses reinforcement learning from calibrated decisions (RLCD) to produce well-calibrated outputs. The conversation explores the need for engineered AI systems composed of multiple specialized components rather than a single monolithic model, and how this shift could reshape agents and tool use in AI-powered software.

This summary was generated from show notes and public descriptions, not from a full transcript review. Details may contain inaccuracies.

Preview

•

Machine-Native Intelligence

Diogo Almeida argues that models optimized for generating text are poorly suited for the kind of fast, reliable decisions required for real-world automation, and that we need 'machine-native intelligence' built specifically for those tasks.

Claude ChatGPT Gemini

•

Reinforcement Learning from Calibrated Decisions (RLCD)

Jev uses a training method called RLCD, which focuses on producing well-calibrated decision-making rather than just maximizing reward, to ensure reliability.

Claude ChatGPT Gemini

1 more ideas & all timestamps

This episode is in its early-access window. The full breakdown unlocks free in about 69 hours — Pro members read everything the moment it lands.

Read it now with Pro$10/mo · founding $96/yr

Was this useful?