Home - TypeSafe AI

Thursday, Nov 12, 2026
reinforcement learning
for calibrated decisions
Version 0.01©2026. All rights reserved.
Made in SF. With Love.
Thursday, Nov 12, 2026
reinforcement learning
for calibrated decisions
Version 0.01©2026. All rights reserved.
Made in SF. With Love.
reinforcement learning
for calibrated decisions
Version 0.01©2026. All rights reserved.
Made in SF. With Love.
Reinforcement Learning from Human Feedback (RLHF) has led to LLMs that are optimized for human preferences. This has led to models that are superhuman at instruction following, and are what we now call “chat.” Yet RLHF creates inherent issues such as mode dropping, overconfidence, and lack of reliability. These flaws mean that LLMs require humans-in-the-loop.
We built a new class of models, System One Models, to be natively used by machines. We’re building with a new architecture, a new sampler, and a new training algorithm: Reinforcement Learning for Calibrated Decisions (RLCD).
Decisions, not strings
Typed outputs that software can act on.
calibrated confidence
Every decision includes an estimate of how confident the model is.
Reliable, fast, and type-safe.
ZmxvYXQgUV9yc3FydCggZmxvYXQgbnVtYmVyICkKewoJbG9uZyBpOwoJZmxvYXQgeDIsIHk7Cgljb25zdCBmbG9hdCB0aHJlZWhhbGZzID0gMS41RjsKCXgyID0gbnVtYmVyICogMC41RjsKCXkgID0gbnVtYmVyOwoJaSAgPSAqICggbG9uZyAqICkgJnk7CglpICA9IDB4NWYzNzU5ZGYgLSAoIGkgPj4gMSApOwoJeSAgPSAqICggZmxvYXQgKiApICZpOwoJeSAgPSB5ICogKCB0aHJlZWhhbGZzIC0gKCB4MiAqIHkgKiB5ICkgKTsKLy8JeSAgPSB5ICogKCB0aHJlZWhhbGZzIC0gKCB4MiAqIHkgKiB5ICkgKTsKCXJldHVybiB5Owp9
*based on workflows for System One tasks (proof)
Watch the real video
Built for automation
Jev returns typed decisions with calibrated probabilities, so your software can account for uncertainty. Set the thresholds for when it acts autonomously and when it asks for review. Combine those decisions in code to build larger workflows, with control over how the intelligence is used.
Workflow Intelligence vs. Cost
Workflow Intelligence vs. Cost
Machine-Native Intelligence
LLMs produce words for people. Jev produces typed decisions and is more like code: reliable, fast, self-consistent, and type-safe.
Every Jev decision comes with a confidence estimate, so your software can act when confidence is high and escalate when it is not.
Per Billion input tokens.
Lower input price than Claude Fable 5.1
Jev: For the builders AND dreamers
TypeSafe A raises Series AI
TL;DR raised a lot of money, and we’re going to make sure it shows up as benefits to y’all 🫶
How Jack & Jill Cut Candidate Screening Costs By 88% With Jev
Reducing cost by 88%, increasing speed 2x, and maintaining quality - all within 10 days from test to prod.
What are System One Models? What is Jev?
System One Models are a new class of AI model built for decisions inside software. Jev is TypeSafe’s first public System One Model, optimized for automation. Send Jev structured questions and get typed decisions with probabilities and confidence that your software can act on.