Directory maps 100+ production JEV architectures: LLMs write, JEV decides, code enforces
alexcovo_eth · x · 2026-09-26
A new directory catalogs 100+ production decision-system (JEV) architectures across routing, context management, browser agents, security, data pipelines and human approvals. Key patterns include:
- route each task to the best model tier
- remove stale tool output from context
- validate risky browser actions
- escalate low-confidence cases to humans
- detect agent-loop stagnation
- verify citations before shipping reports
- score dataset rows before pipeline ingestion
The core architecture: LLMs write → JEV decides → code enforces. The author pairs it with a GPT-6 Astra use-case article, framing a useful agent as many systems with clear jobs.
More from coding & agent
- QuackIR: Jimmy Lin's EMNLP Paper Shows RDBMSes Match Vector DBs for RAG Retrieval — lintool · 2026-09-26
- Where to draw the determinism boundary in LLM pipelines — aronchick · 2026-09-26
- The only pipeline shape that survives production: LLM proposes, rules dispose — aronchick · 2026-09-26
- 100-slot Terminal-Bench 2.1 rerun shows Luna 6 far behind Luna 5.6 — s1lverkin · 2026-09-26
- Qwen3.8-Omni-Flash: natively multimodal agent model with 1M-token context — dair_ai · 2026-09-26
- Que: open-source app queues all your waiting AI agent sessions in one place — Glad_Raspberry_6795 · 2026-09-26