Spike: Jev as a judgement layer to cut model cost
Issue in the Orcha multi-agent orchestration project proposing to move judgment calls onto Jev to cut model cost.
Overview
This entry points to Orcha, a human-authoritative multi-agent orchestration tool in which multiple Claude Code sessions collaborate on an objective through a shared Postgres database while a human approves, reprioritises, reassigns and arbitrates subtasks. It ships as a CLI, a per-project FastAPI and Postgres backing service, slash-command skill templates, a web dashboard and optional desktop app, with a local-first free tier and spend metrics per agent.
The Jev connection is a spike issue described in the directory listing: the project plans to move judgment calls out of its main reasoning model and onto Jev, framed as cutting cost while holding quality. The fetched README does not mention Jev, so this is a proposal rather than a shipped feature.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Other | Orchestration judgment calls (e.g. routing, approval suggestions) moved from the main model to Jev(inferred) | — |
How it works
- Identify judgment calls made by the main model
- Replace them with typed Jev questions
- Compare cost and quality
Strengths
- Potential cost reduction while keeping quality, per the proposal
Caveats
- Planned spike, not implemented in the README
- Jev not mentioned in the fetched README
At a glance
- Type
- Project
- Status
- discussion
- Language
- Python
- Runtime
- CLI + FastAPI/Postgres service
- Has code
- yes
- Has tests
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Issue: a multi-agent orchestration platform plans to move judgment out of its model-of-thought and onto Jev, framed as cutting cost while holding quality.
yibie/awesome-jev
Frequently asked questions
Does Orcha use Jev today?
Not per the fetched README; the listing describes a planned spike.
What is Orcha?
Multi-agent orchestration of Claude Code sessions with human authority over subtasks.
What would Jev do?
Take over judgment calls to reduce cost, per the proposal.
Related projects
Blog post that replaces an agent's classification loop with a single Jev call and reports the loop running 7x faster.
ArticlediscussionX post working through the economics of solving CAPTCHAs with Jev at $0.0068 per hundred versus a market paying a cent each.
DiscussiondiscussionX post listing nine jobs where a typed Jev decision can replace an expensive full LLM call.
DiscussiondiscussionX account of @moritzkremb, cited in a Chinese Jev list as the source of its most specific per-call cost figure.
DiscussiondiscussionExperiment that deliberately misuses Jev as a text generator by asking which symbol comes next and sampling from the returned distribution.
ProjectresearchPythonX post showing a Claude plugin that asks Jev to review redundant tool calls in about a second.
Discussiondiscussion