Agents / mcit

Demand Matching Agent

Agent runGovernancePublicMinimal riskGov Reuse214 runs/30d

Reads an incoming capability request, searches the catalogue and answers whether the capability already exists. Where it does not, it scores the candidate against the national prioritisation criteria and files it in the build backlog, so demand becomes a ranked list rather than a meeting.

Governs the marketplace

Agent runAll governance agents →

An agent executes the process end to end and logs what it did. No human is in the path, because there is no discretion to exercise and no consequence to carry.

Acts for
Marketplace Operator
Action class
advisory
Decisions / 30d
214
Safe-state returns / 30d
1

Processes covered

  • Demand intake and matching
  • Candidate scoring against prioritisation criteria

Stages:discoversteer

Memory

Per-request working memory only; no persistence beyond the demand-gap record it writes.

Safe state

Files the request as unmatched and notifies the marketplace operator when the search index is stale or the rubric returns an error.

Reviewer time per case: 0 min with this agent, 20 without.

Tools

  • Catalogue search index
  • Demand-gap ledger
  • Prioritisation rubric (section 3.7)
  • Investment board writer

Data access scope

  • Read-only: published catalogue metadata and usage metrics
  • Read-only: failed searches and denied access requests
  • Write: demand-gap records and backlog scores only

Guardrails

  • Never marks a capability as existing below 0.8 match confidence, ambiguous matches are listed for the requester to confirm
  • Cannot commit funding or change a backlog rank set by MCIT strategy
  • Requester identity is masked in aggregated demand signals

Human-in-the-loop

  • MCIT strategy forum ranks and commissions the backlog each quarter

Orchestration

Trigger / IntakeLLM Orchestratorfanar-gov-7b-instructCatalogue search indexDemand-gap ledgerPrioritisation rubric (sect…Investment board writerHumansign-off

Every tool call is scoped by the declared data access above; the orchestrator cannot reach systems outside it. Owning entity: MCIT.

Live demo run

Watch the agent execute a real scenario step by step, every tool call, validation, and human checkpoint is traced and auditable. Typical run: ~5.1s.

Owning entity

mcMCIT

Updated 2026-07-26. Run traces retained 24 months for audit under the platform governance policy.

Community · 0 threads

Questions, findings and requests from the entities that use this asset. Owners reply here; threads with upvotes surface to the owning entity's inbox.

No threads yet. Be the first entity to ask, or to share what you found.