Brisora Request access
03 / Policy-controlled decisions

Rules determine what the system is allowed to do.

Every commercial action is evaluated against an explicit, human-issued mandate before it can proceed. The policy engine never infers permission — it checks it.

  • Three outcomes only: allow, deny, or escalate to human review
  • Budget and transaction-value mandates, checked against hard limits
  • Jurisdiction and geography policy, with allow and review lists
  • Counterparty controls, including blocked and watchlisted parties
  • Purchase, payment and contract authority boundaries
  • Margin thresholds evaluated against worst-case outcomes
  • Supplier risk ceilings and required evidence checks
  • Defined stop conditions that halt evaluation outright
04 / Human authority

Automation stops where authority begins.

Decisions that fall outside explicit policy do not default to approval. They route to a human reviewer, and the system is structurally prevented from approving its own proposals. See Principles for the full set of rules this enforces.

  • Explicit escalation for any action a mandate does not clearly allow
  • Self-approval is rejected at the code level — a reviewer cannot approve an action they proposed
  • Fail-closed by default: an unrecognized action class is denied, not permitted
  • Mandates cannot expand their own authority; changes require a new, human-issued mandate version
  • Escalated cases remain open for review rather than auto-resolving
  • Decision context — evidence, rule trace and mandate version — is fixed at the moment of decision

Go deeper

The two capabilities that feed this engine.

01 / Tender intelligence

Structured evidence, not summaries.

Document ingestion, extraction and evidence references — the input side of every decision.

Explore tender intelligence →
02 / Commercial analysis

Commercial viability, evidence by evidence.

Landed-cost, margin and feasibility reasoning — the analysis side of every decision.

Explore commercial analysis →

Built for high-stakes procurement

Built for decisions that need to survive scrutiny.

Brisora is designed for procurement workflows where an unexplained recommendation is not enough.

Complex tenders

Multi-clause requirements with quantities, eligibility conditions and lot structures that need to be resolved, not summarized.

Public procurement

Processes where the basis for a bid or no-bid decision may need to be reconstructed later.

Multi-lot procurement

Participation and feasibility that differ lot by lot, not tender by tender.

Regulated procurement environments

Workflows where authority, budget and counterparty constraints must be explicit and enforced, not assumed.

Supplier-heavy sourcing workflows

Multiple supplier and BOM options that need consistent, evidence-based comparison.

Explicit authority controls

Organizations that already separate who can propose a commercial action from who can approve it.

Current product scope

What Brisora does today.

Tender and RFQ document analysis
Structured evidence extraction
Quantity and requirement extraction
Lot-specific bid feasibility
Participation structure analysis
Supplier and BOM comparison
Landed-cost reasoning
Commercial mandate gates
Procurement policy evaluation
Human review with self-approval protection
Deterministic decision outcomes
Decision provenance and rule-trace audit
Boundary

Brisora is under active development. Some supplier discovery, recurring opportunity discovery and live transaction execution workflows remain outside the current production scope.

External communication and commercial-execution dispatch are deliberately disabled in the current runtime: policy evaluation and review logic run in full, without live dispatch.

See the architecture behind it.

The decision layer that connects evidence, policy and human review runs through six explicit stages.