Some decisions need more than one perspective. Consilium is the decision process, and future service, for assembling specialized AI agents into a panel: architect, security, SRE, strategist, developer, writer and reviewer.
It already exists as a Datarim skill and is being shaped into a reusable product surface for design reviews, production readiness and content councils.
- Status
- In build
- Autonomy (AAL)
- L1 → L3
- Today
- A Datarim skill
- Tomorrow
- A standalone service
01
A fixed protocol
Scope, assemble, analyze, debate, converge, deliver — the same steps every time.
02
Dissent is kept
Failure-mode tables and recorded dissent sit beside the verdict.
03
Roles with viewpoints
Architect, security, SRE, strategist, developer, writer and reviewer each bring their own angle.
Capabilities
Working now
- Panel protocol: scope, assemble, analyze, debate, converge, deliver
- Role-specific viewpoints for architecture and readiness decisions
Status not stated
- Failure-mode tables and dissent capture
- Multi-vendor content council path for voice-sensitive work
- Traceable decision artifacts
Current autonomy level
Weakest link
The process exists, but the standalone service is not productionized; many councils still run as session-local artifacts.
Roadmap to L3
- Persist consilium decisions as first-class artifacts.
- Add reusable panel presets and scoring rubrics.
- Expose a controlled service API for Datarim and Control Arcana.
Next step
Review the current build and its development plan
BUILDING means active development, not a promise of public availability. Use the current capabilities and roadmap above as the reference.
Review the path to the target level