Digital ecosystem · Runtime architecture

A digital architecture built around explicit authority.

The physical process remains local. Digital services organise state, permission, evidence and traceability while local automation and allocated safety functions retain physical authority.

01 Measured input02 Explicit permission03 Local execution04 Traceable outcome

Architecture stack

One connected map with an explicit control boundary.

Commercial services, supervisory governance and plant execution are connected, but they are not interchangeable. The map shows who may advise, who may permit and who physically acts.

Runtime at a glance

Inputs become a request. A verified request becomes local work.

Inputs, digital gates, execution authority and recorded outputs remain visibly separated throughout the decision path.

Inputs

Observed & configured

  • Laboratory and sensor data
  • Device and module state
  • Kits, recipes and envelopes
  • Operator identity and request
  • Approved external-system context
Digital gatesState→Route→Rules→PermissionUnknown, stale or conflicting evidence is not implicit approval.
Local execution & outputs

Acted & recorded

  • Released PLC sequence
  • Asset operation and protective response
  • Time-stamped data and events
  • Alarm and operator acknowledgement
  • Outcome evidence and engineering report
IDIdentity & access
DATAData service
CFGConfiguration
AUDAudit record
APIIntegration
TIMETime service
Runtime overview

Eight cross-cutting orchestration roles.

The overview uses eight broad roles. The detailed architecture below separates the same work into fifteen named service engines.

Runtime roleFunctional responsibilityAuthority boundary
FRNEStructure and compare feed condition against the configured referenceRequires qualified measurement and a released envelope
StateMaintain the current configured state of modules, equipment, routes and materialState record is not physical confirmation unless linked evidence supports it
RouteEvaluate the intended path, lineup, isolation and resource availabilityLocal feedback and interlocks remain authoritative
RulesApply configured constraints, limits and applicable operating rulesCannot override allocated safety functions
ExecutionSequence a permitted request for release to local automationPLC executes; cloud and advisory services do not actuate
SafetyEvaluate digital permissives and monitor the status of allocated protectionsSIS / ESD retains its allocated veto and de-energise authority
ConvergenceEvaluate stability and movement toward a target using an approved basisNo predictive result is accepted without a qualified model or test basis
AlarmClassify events, preserve context and present required interventionOperators and site procedures govern accountable response

Decision progression

A process step earns the right to proceed.

A failed gate pauses progression and directs the process to a defined correction, rejection, safe state or accountable operator decision.

S0 SafeS1 PrepareS2 PrimeS3 RunS4 Abnormal / tripS5 Service / recovery

Operative runtime architecture

Fifteen named engines, grouped by control function.

The engine set is grouped by feed and equipment truth, configuration, authorisation, evidence and fleet support. Names describe intended responsibilities; project implementation still requires released software, interfaces and test evidence.

A

Feed & equipment truth

3 engines
  1. 01
    Feed Reference Neutralisation

    Structures feed measurements, compares them with the configured reference and records required correction or review.

  2. 03
    Sensor Truth

    Reconciles available signals and records freshness, disagreement, drift and the evidence used to accept a value.

  3. 04
    Valve Truth

    Compares commanded position with available position, flow and pressure feedback.

B

Configuration & routing

4 engines
  1. 02
    Kit Envelope

    Holds the released operating window and kit-specific limits for the selected application.

  2. 05
    Liquid Routing Truth

    Checks the requested path against configured valves, isolation, flush and recirculation routes.

  3. 06
    Module State

    Maintains named module states such as prepare, run, pause, clean, isolate and fault.

  4. 07
    Setpoint Translation

    Translates an authorised correction into a bounded setpoint request for local automation.

C

Safety & authorisation

3 engines
  1. 08
    Safety Permission

    Checks the digital permissive record while PLC and independent safety functions retain their allocated authority.

  2. 10
    Alarm & Intervention

    Classifies events and presents the operator with the available context and required response.

  3. 11
    Operator Authorisation

    Associates identity and role with the set of requests an operator may submit or acknowledge.

D

Evidence & lifecycle

3 engines
  1. 09
    CIP / Cleaning Trigger

    Records when cleaning is due and the evidence required before a subsequent configuration is released.

  2. 12
    Digital Thread

    Links source, configuration, request, decision, execution record and available outcome evidence.

  3. 15
    Olerib Verification

    Records the policy and evidence considered when a request is permitted, denied or escalated.

E

Fleet & support

2 engines
  1. 13
    Fleet Data

    Organizes approved operating history for comparison, service and engineering review across deployed units.

  2. 14
    Remote Operations

    Supports supervised remote assistance within the site's access, cyber-security and authority rules.

FRNE · Feed Reference Neutralisation Engine

Turn measured feed reality into an explicit release decision.

FRNE is presented as a deterministic feed-reference workflow. It does not infer an approved process envelope: that envelope must come from qualified project work.

01MeasureLab, sensors and field records
02TokeniseSource, unit, quality and time
03EvaluateCompare with the released envelope
04RecommendPropose a bounded correction
05PredictCheck response against an approved model
06ReleaseIssue a setpoint request only when permitted
07ExecuteLocal automation performs the action
08VerifyNew evidence re-enters the loop
Examples of detected deviation
  • Solids or viscosity outside the released window
  • pH, ORP, conductivity or density outside limits
  • Unexpected feed drift or signal disagreement
  • Route, equipment or module state not confirmed
Candidate response—not an automatic command
  • Dilute, blend, heat, cool or recirculate
  • Pause for laboratory or operator review
  • Reject the feed or isolate the route
  • Trigger cleaning review before configuration change

Olerib Foresight · Advisory runtime

Evaluate collectively. Judge locally.

Foresight evaluates one questioned signal or proposed action inside the available approved context and projects its expected governability across three evidence horizons. It returns a bounded recommendation to Olerib Vision—not permission and never a plant command.

3 × 3STEP 1STEP 2STEP 3
Signal 1Expected evidenceExpected evidenceExpected evidence
Signal 2Expected evidenceExpected evidenceExpected evidence
Signal 3Expected evidenceExpected evidenceExpected evidence
BOUND OUTPUT

One advisory recommendation

  • Questioned signal or action
  • Suggestion source and collective context
  • Step 1 / 2 / 3 projection
  • Current path versus proposed path
  • Governability and superiority result
  • Recoverability, interruptibility and collapse conditions
  • Confidence result and witness record

Token language

Eight token families make the decision inspectable.

A token is a structured record, not a physical asset or a performance claim. Together the families describe what was observed, requested, permitted, executed and verified.

F

Feed state

Measured feed value with unit, source, quality and time.

A

Action

Proposed bounded correction or operational action.

SP

Setpoint

Released target linked to the evidence and action that justified it.

V

Verification

Available sensor, laboratory or reconciliation evidence.

R

Route

Configured material path, including isolation, flush and recirculation.

S

Signal quality

Freshness, agreement and status of the contributing measurement.

M

Module state

Named operating state visible to the relevant runtime functions.

P

Policy

Permission, role and boundary applicable to the requested action.

StateRequired evidenceSetpoint statusFeed progression
Outside envelopeOne or more accepted parameters outside released limitsBlockedBlocked
Correction plannedBounded response proposed; authority and model basis openPendingBlocked
ConvergingApproved model or test basis predicts entry into target windowPendingBlocked
Neutral / in windowAccepted parameters inside target window and stable for the specified criterionReady for reviewProject inputs required
ReleasedRequired policy, route, state and verification gates satisfiedIssuedReleased to local execution
MaintainingMonitoring remains valid inside the released operating windowActiveContinue with monitoring

Digital ecosystem products

Eight services around one local plant authority.

Information can move across the ecosystem; command authority does not move with it. Each service has a named role and an explicit boundary.

Olerib Vision — The Architecture of Intent
Independent governance IP

Olerib Vision originated with Adapt but remains a standalone product that can be integrated with third-party industrial environments.

Understand Olerib Vision →
ServicePrimary roleInformation handledAuthority boundary
ADI / Adapt RadarMarket and deployment intelligenceFeedstock and tailings mapping, partner and opportunity research, regional demand contextNo process authority
Adapt ExchangeIndustrial collaboration layerFeedstock and capacity coordination, solution and kit collaborationNo process authority
Adapt ConnectConnected data backboneTelemetry, fleet visibility, configuration delivery, maintenance signals and supervised remote supportTransport only; local authority remains
Adapt Intelligence / Fleet IntelligenceAnalytics layerDiagnostics, unit and hub comparison, capacity context and advisory forecastingRecommendation only
Adapt HMIOperator interfaceState, alarm, permissive and operating-window presentationOperator requests; PLC executes
Olerib VisionGovernance and witnessPolicy evaluation, evidence context and allow / deny decisionNo final-element actuation
Olerib ForesightBounded advisory engineCollective context, three evidence horizons, governability comparison and a traceable recommendationAdvice only; Olerib independently decides
AI / deployment advisoryConcept-scope advisory servicePattern detection, scenario support and summaries using governed dataNo permission or execution authority

Intended operational value

Benefits must be earned by implementation and evidence.

Each intended outcome is paired with the evidence needed before it can become a public performance claim.

Intended outcomeArchitecture contributionEvidence required before claiming performance
More stable qualified feedExplicit feed envelope, deviation handling and release stateRepresentative feed campaign, acceptance criteria and trend data
Lower transition riskNamed route, state, permission and cleaning gatesHAZOP / risk review, cause-and-effect, FAT / SAT and operating procedures
Operator accountabilityIdentity, role, acknowledgement and linked decision recordAccess-control design, audit policy and record-retention verification
Faster troubleshootingSource-to-outcome digital thread and time-correlated eventsCommissioned instrumentation, time synchronisation and service records
Repeatable commissioningControlled configuration, recipes and acceptance gatesApproved commissioning plan and comparable site evidence
Fleet comparisonConsistent data definitions and approved cross-unit recordsData-quality rules, comparable operating context and governance approval
Reduced dependence on tacit knowledgeReleased procedures, configured envelopes and visible decision contextTraining, competency assessment and controlled-document review
AI-ready analyticsStructured, traceable data for advisory modelsModel validation, monitoring and a documented non-actuation boundary

Control boundary

Credibility begins with explicit authority and evidence boundaries.

Deterministic target

Local execution

Released sequences and interlocks run at the plant, not through an internet-dependent control loop.

Supervisory

Olerib permission

Policy and evidence can permit, deny or escalate a request without directly driving a valve or motor.

Evidence required

Formal claims

No public SIL level, zero-latency, immutability, cyber-security or universal performance claim is made without a released conformity basis.

From architecture to project

Define the feed, route and evidence before configuring the digital layer.

Navigator creates a preliminary functional flowsheet and records the missing project inputs. The digital architecture is then engineered around the selected process and authority model.

Open Navigator →