Use casesAutomation
Automation builds keep Jev in the loop — controlling agents, gating steps, and choosing what happens next as state changes.
Jev acts as a decision layer in voice pipelines, utilizing typed primitives like Noul (yes/no with probability), Choice (labeled decision), and Score (scalar on a defined rubric) to process conversational turns before and during LLM responses.
This project enables computer use within a Dockerized Linux environment by using the accessibility tree as 'eyes' and TypeSafe Jev to answer typed questions for the next action, with a11y invoke or xdotool serving as 'hands'.
Jev is an AI model that provides typed decisions, probabilities, and confidence scores for software to act upon, differing from traditional LLMs by not generating answers token by token.
This project is a browser agent where Jev makes all decisions and an agent-browser acts.
Jev is an AI system that makes typed decisions and probabilities based on state, contrasting with traditional AI models that generate text.
Jev is a new System One model from TypeSafe AI designed to accelerate decision-making in agentic systems by processing state and typed questions to produce calibrated decisions in under a second, eliminating the need for text parsing.
This project integrates AI into an agent-based city model, replacing half of the property developers with AI agents that independently decide land bids.
This project tests JEV within Automata Room to control a robotic hand, enabling finger-by-finger manipulation and the creation of shareable robot episodes.
This project integrates Jev and GPT-6 Astra on an Agilex PiPER robotic arm, with a process involving model skill selection, governor checking, and arm movement.
Jev is an AI model developed by TypeSafe AI that is trained for making decisions using Reinforcement Learning from Human Feedback (RLHF) and Reinforcement Learning from AI Feedback (RLAIF) or similar techniques, rather than generating text. It is designed for high-frequency, low-cost decision-making in agentic systems.
Jevton is a simulated town where an AI named Jev makes all decisions for 120 residents, including their actions, work, spending, health, and community priorities, processing over 20,000 decisions in under 5 minutes.
This project demonstrates Jev acting as a computer by emulating the Motorola 6502 processor, allowing it to run short programs directly.
A live trading lab for Kalshi, powered by Jev, that processes BTC and ETH market data to make typed decisions and enforce risk limits.
Flysim is a simulation where a fruit-fly navigates a foraging world, seeking sugar and evading a predator, with the ability to swap decision-makers.
Entropy City is a simulation where an AI named Jev coordinates disaster response and recovery decisions for a city experiencing cascading failures.
An autonomous drone autopilot that navigates from point A to point B while avoiding obstacles.
Marionette interacts with a Flutter game by reading its widget tree and tapping elements, while Jev makes decisions on how to interact.
Jev improvises contemporary dance moves using its head, arms, hips, legs, and footwork, with each body part having its own target and timing, driven by a local motion engine and a prediction model.
