AMR · Policy → Evidenz → Kapital

Trace what is funded. Verify what is executed. Identify what matters next.

AMR decisions depend on evidence distributed across national plans, surveillance systems, research, donor portfolios, projects and implementation records. AgentBrain reconstructs those signals into a decision-grade intelligence layer.

AgentBrain AMR decision intelligence surface showing memory, validity and evidence state
Decision chain: policy, evidence, capital, execution and outcome boundaries
Entscheidungskette

Turns scattered AMR evidence into a bounded decision chain.

The system does not treat every related health programme as AMR funding. It separates direct AMR capital, linked portfolios, enabling exposure, wider health-financing context and unresolved attribution.

PolicyCommitmentsPlans, priorities, responsibilities and implementation structures.
EvidenceObserved signalsSurveillance, research, representativeness and quality.
CapitalAttributionDirect, linked, enabling, context or unresolved financing.
ExecutionDelivery proofProjects, components, procurement, disbursement and implementation records.
OutcomeBounded linksResults are connected only where evidence supports the relation.
Substrate-Beitrag

Wie das Substrate die Analyse verändert.

Memory

Keep evidence history intact

Country plans, funder signals, outreach, prior claims and negative findings remain inspectable instead of disappearing between analysis sessions.

Validity

Downgrade weak claims

No claim can be stronger than its source, scope and transformation path. Related records do not automatically become verified relationships.

Prioritization

Rank what to learn next

Unknowns are evaluated by decision impact, information gain, portfolio sensitivity, network effects and resolution cost.

Alpha agent

The AMR use case becomes concrete through Alpha.

Alpha is the research agent inside the AMR intelligence layer. It turns a country, funding, programme or implementation question into evidence work that can be reviewed: sources, claims, gaps, confidence and boundaries stay visible.

AskDecision questionCountry, portfolio, funder, programme or implementation problem.
TracePublic evidencePolicy, surveillance, capital, research and implementation records.
ReviewClaim disciplineSupported, partial and unknown states remain separate.
PackageIntelligence packBrief, evidence ledger, funding map or source/claim matrix.
GateHuman approvalPublic output remains reviewed and controlled.

Use Alpha when the question is operational. The parent AMR page explains the decision chain. Alpha explains the actual work a reviewer can ask the agent to perform.

Open Alpha use case

Where it helps

Use AMR Decision Intelligence when funding visibility, implementation evidence and surveillance signals are being confused. The output is not another dataset; it is a structured account of what is known, what is connected, what remains uncertain and what should be investigated next.

Where proof stops

Not medical

No clinical or medical decision is delegated to the system.

No overclaim

Missing public evidence does not prove missing national activity.

Human review

AMR outputs remain subject to source review and methodological scrutiny.

The AMR surface is live as a separate vertical landing page. It contains the request form and the proof-stack entry point for deeper inspection.

Visit amr.agentbrain.ch
Verified sources: live amr.agentbrain.ch page, AMR proof-stack page, local AMR proof-run artifacts under AlphaAgent/proof-videos, and TheshBrain memories 85a11955, 3cd1919e, 326b0de8.