Home · Work · System design
Data and AI platformPlatform lifecycle and release controllayered

Retrieval-quality and provenance monitor

High recall can coexist with stale, unauthorised or semantically misleading evidence that produces a fluent but unsafe answer.

Approach

The monitor evaluates retrieval by authority, applicability, freshness, citation entailment and route outcome, not relevance alone.

Organisations, systems and operating conditions are intentionally anonymised and recomposed. The design demonstrates engineering and banking-domain reasoning; it does not represent a named client estate, vendor product or measured production result.

Overview

System overview

Five operating planes separate interaction, identity, decision control, authoritative state and operating evidence.

Retrieval-quality and provenance monitor: logical system architecture A five-plane architecture showing channels, access controls, twelve platform services, authoritative domain systems, evidence and operations. Solid nodes are active for this illustrative case; dashed nodes are optional platform capabilities. aSystem boundary and decision authoritySchematic architecture · illustrative scenario Experience planeAccess and scope planeDecision control planeAuthoritative data and effectsEvidence and operations rail Customer Channelweb · mobile · voiceStaff Decision Workspacereview · explain · approveOperations Control Roomincident · replay · withdrawRequest Ingress GatewayTLS · rate limit · schemaIdentity & Purpose GateOIDC · subject · consentPayload MinimiserPII policy · field maskCapability Directoryversion · scope · healthEvidence ContextCompilerContext & evidenceParty RelationshipResolverIdentity & relationshipsAuthority Rules EngineAuthority & policyJourney StateCoordinatorWorkflow & custodyTyped Action BrokerIntegration & effectsDecision ReceiptLedgerEvidence & auditReasoning RouteManagerModel executionEffective-DatedKnowledge HubKnowledge & rulesBehaviour EvaluationHarnessEvaluation & releaseOutcome RecoverySupervisorResilience & recoveryOperational Read ModelWorld state & read modelsDecision ReviewWorkbenchHuman decision & operationsAgentauthoritative recordtool registryauthoritative recordmodel gatewayauthoritative recordknowledge servingauthoritative recordworkflow runtimeauthoritative record HTTPSHTTPSoperations context Immutable Evidence ArchiveTrace · Metric · Cost PipelineIncident · Replay · Custody Queue telemetryexceptions executed or authoritative pathproposal or optional capabilityevidence and telemetryactive in this case
Figure 1. The system boundary separates interaction, authority, business state and operating evidence. Solid services participate in this case; dashed services remain outside the admitted route. This is an illustrative system model, not a named deployment.

Permitted workThe platform supplies reusable controls and contracts. Domain teams retain business logic, data purpose and outcome ownership.

Consistency ruleTie deployed code, prompt, model, data, permissions, policy and evidence to one release manifest and compatibility graph.

Hard boundaryThe model is not a system of record, identity provider, policy authority or proof that an external effect occurred.

Deployment

Deployment

Agents, prompts, tools, policies and models must be built, certified, operated, changed and retired as one reachable system.

Retrieval-quality and provenance monitor: deployment and trust-zone topology A production topology separates public edge, identity boundary, workload cluster, integration plane, data enclave and operating controls. Connections name transport and identity protocols, and a second region provides controlled failover. bDeployment topology and failure domainsLogical deployment · no measured estate data EDGE AND ACCESS PRIVATE WORKLOAD CLUSTER INTEGRATION AND DATA ENCLAVE OPERATIONS, EVIDENCE AND RECOVERY PLANE Web applicationfirewallTLS · bot · rate API ingressschema · quota Identity brokerOIDC · MFA Purpose gateclaims · consent HTTPStokensubject Regional traffic managerhealth · canary · failover Workload identity issuershort-lived service tokens Event & work queuepartition · DLQ · backpressure Secrets & key serviceenvelope encryption · rotation Evidence ContextCompilerzonal workloadAuthority RulesEnginezonal workloadDecision ReceiptLedgerzonal workloadEffective-DatedKnowledge Hubzonal workloadBehaviourEvaluation…zonal workloadReserved workloadslotdisabledReserved workloadslotdisabledReserved workloadslotdisabled HTTPS · scoped JWT workload token async eventkey lookup IntegrationgatewaymTLSAgentsystem of recordtool registrysystem of recordmodel gatewaysystem of recordknowledge servingsystem of record Model route enclavepolicy · schema · egress Qualified data storesSQL · graph · object Trace, metric & cost pipelineOpenTelemetry · redaction Decision receipt archiveappend-only · retention Reconciliation workersreadback · replay · DLQ On-call & review queuescustody · SLA · escalation Secondary-region standbycompatible manifest telemetryreceiptsexceptions
Figure 2. Workloads remain isolated from systems of record. Short-lived identity, typed integration contracts, append-only evidence and controlled failover define the deployable boundary; the topology is schematic.
Authority checkpoint

Content failing provenance or entitlement checks cannot enter the model context, regardless of similarity score.

Runtime

Runtime flow

A deterministic outer workflow contains model-led work inside typed, observable calls. Dashed messages remain proposals until policy or a human grants authority.

Retrieval-quality and provenance monitor: build, release and retirement lifecycle A controlled lifecycle links registration, design, implementation, evaluation, approval, canary release, operation and retirement. A signed manifest binds dependencies and evidence; failed gates and operating drift return to an owned change route. cRuntime protocol and control pointsProtocol model · illustrative execution Build and releaseOperate, change and retireManifest, evidence and recovery Signed system manifestowner · code · model · knowledge · tools ·… Registerowner · purpose · data boundaryDesigntopology · authority · contractsBuildcode · prompts · rules ·…Evaluatefunction · failure · harm · NFRApprovesigned manifest · evidenceCanarylimited traffic · rollback…Operatebehaviour · cost · incidentsRetirerevoke · archive · prove closure Evaluation evidenceroute · failure · compatibilityRelease registrysignature · canary · rollbackOperating evidencetrace · cost · outcomeIncident & change queueowner · severity · deadlineArchivewithdrawal proof Gate rule: Automate evidence collection before automating approval. Expand self-service only when the paved route proves safer and faster than bespoke delivery.
Figure 3. Design, evaluation, release, operation and retirement remain bound to one signed manifest. A failed gate or operating incident returns to an owned change route. The paths are protocol semantics, not an observed production trace.
S

State model

A signed system manifest links code, prompt, model, knowledge, tools, permissions, policies, tests and owner. Any material dependency change creates a new release candidate.

I

Integration choice

Build and runtime boundaries use the same schemas. Registry metadata drives discovery, policy, telemetry attribution and deprecation.

C

Consistency semantics

The deployed artefact, registry entry and evidence pack share one immutable release identifier; partial promotion is not a valid state.

Services

Services and interfaces

These roles are deliberately vendor-neutral. Each can be independently owned, versioned and replaced.

  1. CAS
    Context and evidence

    Evidence Context Compiler

    Accepts a typed evidence request and returns the minimum permitted facts with source, event time, observation time, validity and exclusions.

    Degraded routeExclude stale or unauthorised fields, ask a bounded question, or stop the route. Never fill a gap with an inferred fact.
  2. DES
    Evidence and audit

    Decision Receipt Ledger

    Appends request, versions, policy result, model proposal, approval, action receipt, readback, correction and custody events under one correlation key.

    Degraded routeBlock a consequential close when mandatory evidence is missing; keep the case open with an explicit evidence defect.
  3. CKR
    Knowledge and rules

    Effective-Dated Knowledge Hub

    Serves owned, audience-qualified and effective-dated content; exposes supersession, withdrawal and dependency metadata.

    Degraded routeServe the last approved version only when policy permits; otherwise surface conflict or unavailability to the owner.
  4. SEL
    Evaluation and release

    Behaviour Evaluation Harness

    Runs component, route, trajectory, failure, harm and outcome tests against the versioned system manifest.

    Degraded routeReduce or withdraw authority when a dependency changes or a critical suite fails; preserve the last approved route where compatible.
  5. PDS
    Authority and policy

    Authority Rules Engine

    Evaluates identity, purpose, capability, amount, risk tier and policy version; returns allow, deny, step-up or human-review with reasons.

    Degraded routeFail closed for effects and restricted data. Read-only explanation may continue only from approved public or customer-visible sources.
Figure 4. Each service exposes a typed contract and a defined degraded route. Coordination stays in the control plane while domain ownership remains outside it.
Data

Data and records

Durable records carry provenance, authority, effect and custody without turning a transcript into an uncontrolled memory store.

Retrieval-quality and provenance monitor: state ownership and evidence lineage Authoritative source observations feed six versioned runtime records. Each record names a single write owner. An append-only event and evidence stream builds a disposable current-state projection and supports audit replay. dState lineage, ownership and temporal validityLogical records · fields are illustrative AUTHORITATIVE INPUTS · POINT-IN-TIME QUALIFIEDAgentsource version · observed attool registrysource version · observed atmodel gatewaysource version · observed atknowledge servingsource version · observed atqualified inputqualified inputqualified inputqualified input WRITE OWNER · Evidence Context CompilerRequest enveloperequest_idpurposeactor_refsubject_refversioned · attributable · retainedWRITE OWNER · Authority Rules EngineContext manifestcontext_idsource_refsource_versionobserved_atversioned · attributable · retainedWRITE OWNER · Typed Action BrokerCase statecase_idstatestate_versioncurrent_custodianversioned · attributable · retainedWRITE OWNER · Reasoning Route ManagerAuthority decisiondecision_idpolicy_versionrequested_capabilityoutcomeversioned · attributable · retainedWRITE OWNER · Behaviour Evaluation HarnessAction and effect receiptaction_ididempotency_keyinput_hashdispatch_statusversioned · attributable · retainedWRITE OWNER · Operational Read ModelCustody eventtransfer_idfrom_ownerto_ownerreasonversioned · attributable · retained APPEND-ONLY CASE EVENTS + DECISION RECEIPTSrequest hash · source versions · policy version · proposal · authority · effect readback · custody Operational current-state viewrebuildable projection
Figure 5. The record chain preserves source version, write ownership, authority and custody. Solid records are required in this scenario; dashed records remain compatible but dormant.
Consistency contractDomain source rule

Tie deployed code, prompt, model, data, permissions, policy and evidence to one release manifest and compatibility graph.

Choices

Trade-offs

The selected design is not universally superior. It is the safer fit for this boundary and failure cost.

  1. 01

    Prefetch context or acquire it when needed

    Selected path

    Prefetch stable, purpose-safe facts; acquire volatile facts on demand against a time-qualified snapshot.

    Rejected path

    Load a broad customer profile at session start.

    Cost acceptedThe selected design adds source calls and latency, but reduces stale data, excess exposure and accidental reuse.

  2. 02

    Mutable current-state record or append-only decision history

    Selected path

    Use append-only events plus a rebuildable current-state projection.

    Rejected path

    Overwrite the case row with its latest status.

    Cost acceptedReplay and storage are more complex, but point-in-time reconstruction and correction lineage remain possible.

  3. 03

    Centralise all content or federate ownership behind a shared contract

    Selected path

    Federate authoring while centralising lifecycle metadata, validation and serving rules.

    Rejected path

    Create one centrally authored knowledge corpus.

    Cost acceptedFederation requires stronger contracts and owner discipline, but preserves domain accountability and release velocity.

  4. 04

    Evaluate model responses alone or certify the complete reachable system

    Selected path

    Test prompts, models, tools, knowledge, policies, state transitions and human paths together.

    Rejected path

    Use a static answer-quality benchmark as the release gate.

    Cost acceptedSystem evaluation takes longer and needs synthetic environments, but detects authority and recovery failures that answer scoring misses.

  5. 05

    Reason over policy text at runtime or compile executable rules

    Selected path

    Compile stable decision logic and retain retrieval for explanation and residual ambiguity.

    Rejected path

    Ask a model to interpret the source document for every request.

    Cost acceptedRule compilation needs controlled change, but creates repeatable decisions, regression tests and clear exceptions.

Figure 6. Teal marks the selected route; coral dotted rules preserve the rejected alternative. Each decision states the cost accepted rather than presenting one architecture as universally superior.
Recovery

Failures and recovery

Retries are bounded by knowledge of business effect; unknown outcome remains visible, owned and independently reconciled.

Retrieval-quality and provenance monitor: failure classification and recovery control flow Five failure sources converge on a classifier that distinguishes known no effect, known failure, unknown outcome, material evidence gaps and control-plane failure. Each state has an explicit recovery route, service degradation and evidence receipt. eFailure classification and recovery policyControl model · thresholds set by owners DETECTION POINTSEFFECT CLASSIFIERRECOVERY DECISIONSERVICE-SPECIFIC DEGRADATION Source dependencyA highly ranked passage can be accurate in…Authority servicePolicy is unavailable, expired, or returns…Reasoning routeOutput fails schema, citation, grounding…Action pathTransport times out after dispatch & the…Evidence railReceipt or custody append fails after a… Outcome classifierattempt · idempotency · source readback ·… Known no effectSafe bounded retry with the same…Known failureNarrow, queue, compensate, or fail closed…Unknown outcomeRead back authoritative state before retry…Material evidence gapRetain custody & create an owned…Control-plane failureStop promotion or dispatch; restore the…Evidence Context CompilerExclude stale or unauthorisedfields, ask a bounded question, orstop the route. Never fill a gap…Decision Receipt LedgerBlock a consequential close whenmandatory evidence is missing;keep the case open with an…Effective-Dated Knowledge HubServe the last approved versiononly when policy permits;otherwise surface conflict or…Behaviour Evaluation HarnessReduce or withdraw authority whena dependency changes or a criticalsuite fails; preserve the last…Authority Rules EngineFail closed for effects &restricted data. Read-onlyexplanation may continue only from… RECOVERY RECEIPTfault origin · classification basis · owner · retry/compensation key · readback reference · residual risk · resolved at classifier evidencerecovery evidence Pattern rule: Canary, kill switch& rollback restore the lastcompatible manifest…
Figure 7. Recovery follows the known business-effect state, not the transport symptom. Every branch ends in a receipt preserving fault origin, custody and reconciliation basis.
Capacity

Performance and capacity

Actual thresholds belong to accountable service owners. The design exposes the equations and observables that those owners must baseline.

Dependent variablesOperating
envelope
  • deployment lead time
  • critical test coverage
  • rollback time
  • cost and trace attribution
Capacity relationships
  1. 01gate_runtime = test_cases x routes x dependency_versions
  2. 02release_capacity = available_environments / average_gate_duration
  3. 03operating_cost = model + tools + platform + human_review + incidents
Privacy lens

Enforce purpose, tenant and workload identity at every hop; telemetry stores metadata rather than unrestricted prompts or payloads.

Figure 8. The variables define what must be measured before capacity or quality claims can be accepted. Relationships are analytical scaffolds, not invented targets or production results.
Operations

Release and operations

A design is production-ready only when teams can prove what happened, recover it and change it safely.

Minimum release gates
  1. 01

    reachable-authority analysis

  2. 02

    cross-version compatibility

  3. 03

    failure and rollback rehearsal

  4. 04

    cost, latency and trace attribution

Figure 9. Promotion requires evidence at each gate and a compatible rollback route. No gate is implied to have passed; accountable owners set thresholds and sign the record.
Operating stateReleaseRecover
Evidence

Query set, candidate corpus, filter decisions, ranking trace, entailment checks, answer outcome and drift alert.

Human authority

Knowledge and model owners set thresholds, investigate drift and approve corpus changes.

Evolution

Automate evidence collection before automating approval. Expand self-service only when the paved route proves safer and faster than bespoke delivery.

Ownership

Platform, domain product, data, security, model-risk and service owners share certification decisions.

Figure 10. Evidence, human authority, ownership and controlled evolution remain coupled throughout operation. Removing any quadrant breaks the assurance case.