Docs
OfficialJev Practices & Patterns
How to Build with TypeSafe
Design AI-powered software by keeping code in control and giving System One narrow, structured decisions to make.
- Format
- Docs
- Reading time
- 2 min
- Published by
- TypeSafe AI
- Last verified
- Sep 2026
Used inJev for Software & DevTools
Highlights
- Most queries complete in about 100 ms, fast enough for real-time request paths and user interfaces.
- The guide targets a greater than 100x intelligence-to-speed-and-cost ratio for System One.
- The spam example replaces one broad 'Is this spam?' Noul with six atomic Nouls such as requests_credentials and sender_identity_mismatch.
- Choice criteria can be objects with what, not_for, and examples fields so easily confused options stay distinct.
- Code composes answers with weights, e.g. 0.45 x requests_credentials + 0.30 x sender_identity_mismatch + 0.25 x unexpected_reward.
Quickstart
python
answers = response.answers
quality = (
0.4 * answers["answers_request"].noul
+ 0.4 * answers["citations_are_supported"].noul
+ 0.2 * (1 - answers["contradicts_context"].noul)
)Watch out
System One does not generate code or choose its own next action, so agent-style loops should be replaced by software workflows that call the model only for narrow, typed decisions.
Questions about this entry
Short answers drawn from the linked project or page.
Most queries complete in about 100 ms, per the guide, which is why it targets real-time request paths and user interfaces and cites a greater-than-100x intelligence-to-speed-and-cost ratio.
As six atomic Nouls — requests_credentials, sender_identity_mismatch, unexpected_reward, and so on — rather than one broad 'Is this spam?' question, with code composing the answers, for example 0.45 x requests_credentials + 0.30 x sender_identity_mismatch + 0.25 x unexpected_reward.
No. It does not generate code or choose its own next step, so your code owns the workflow and the model only handles narrow, structured decisions.
Use a backticked dot-and-index path inside the question, such as support.tickets[0].message, so the question targets one specific nested value.
Include only the context relevant to the current questions; the guide says this helps the model avoid distractions and context rot, and warns not to rely on knowledge stored in model weights.
Greater than 100x, on the bet that cheaper intelligence will create much more demand.
Traditional software as a complex decision tree, LLM agents that choose their own steps, and AI-powered software where code owns the control flow and the model makes atomic, constrained decisions.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: How the agent stack moved in Jev's first three days
YouTube: Diogo Almeida tells swyx what System One is, why RLHF failed his automation goal, and why AI should eventually disappear.
X: The 10-step roadmap, summarised
X: Giving your agents a decision brain
More Jev resources like this
Instructions, option descriptions, score levels, and noul criteria all accept JSON, so you can label question parts, pass schemas and taxonomies, and keep candidate paths alive.
Jev Practices & PatternsDocs#official#primitives#architecture
Official
Architectural patterns for System One systems: speculative fan-out, confidence-gated routing, composite scoring, and intent routing.
Jev Practices & PatternsDocs#official#patterns#architecture
Official
The box question type: one bounding box per question, returned as four edges that are fractions of the image, plus confidence, with no criteria field and no not-found value.
Jev Practices & PatternsDocs#official#docs#vision
Official
A maintained list of jev-1.13's known failure modes, literal reading, unreliable counting, dates as text, indirection, context rot, and contradictory criteria, each with a guardrail.
Jev Practices & PatternsDocs#official#docs#evaluation
Official
Break a judgment into atomic Score questions, normalize each by its top level, and combine them with weights you control in code, as in the resume example's four scored dimensions.
Jev Practices & PatternsDocs#official#patterns#score
Official
Use the answer to decide what to do and confidence to decide whether to act: the voice-banking example routes below 0.6 to a human and needs 0.85 or more to auto-approve a transfer.
Jev Practices & PatternsDocs#official#patterns#confidence
Official
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.