Public proof surface

Evidence reconciled through 20 July 2026

Release scope: Architecture, Achievements, Knowledge Sharing

Source: GPT KB + Git

Curated static release — not a continuous live-status feed

Release: AIOS profile v0.2 + Governance layer update

← Back to AIOS Case Study Map

Org Chart & Role Policy

Public-safe operating map for who owns executive orchestration, execution control, specialist delivery, evidence review, and budget discipline inside the AIOS work system.

This is an operating model summary, not runtime access control, autonomous staffing, or production-readiness evidence.

Capability Detail Pages

These compact cards open deeper role capability pages. Long learning, evidence, and upskill narratives stay off the hub so the role map remains scannable.

Production readiness control layerDraft public-safe slice; taxonomy not PASS/DONE/VERIFIED
Big Crew Architecture Gate

Keeps prototype-to-production decisions bounded, evidence-led, and public-safe before source work expands.

Architecture gateClaim boundaryProduction slice
View capability detail
Big Crew v0.1 quality roleChecker Designer is a capability, not a separate role
QA Sentinel

Protects evidence sufficiency, validation readiness, claim discipline, and release or handoff quality boundaries.

Evidence sufficiencyValidation readinessChecker Designer capability
View capability detail
Telemetry capability layerNo production telemetry proof by default
Data Team

Turns dashboard-shaped telemetry into evidence-first validation and claim-safe insight.

Telemetry semanticsData qualityInsight readiness
View capability detail
Execution acceleration layerNot Gate or approval authority
Runner / Super Runner

Executes scoped work inside approved boundaries while preserving handoff, evidence, and escalation discipline.

Scoped executionEvidence captureEscalation routing
View capability detail
Repository integrity capabilityProduction evidence: NOT_PROVEN
Repo Custodian

Protects source custody, exact-SHA work, recoverability, and repository evidence without turning local proof into production claims.

Source custodyWorktree isolationRecovery readiness
View capability detail

Operating Org Chart

KB-sourced summary

Lyn

Owner / Decision Authority

Robert / GPT

Executive Orchestration Layer

Execution Control & Evidence Governance

Scope, authorization, validation, stop conditions, and claim boundaries

Super Runner

Execution Control Layer

Runner

Bounded Task Executor

Checker

Evidence & Claim Boundary Checker

Big Crew

Specialist Delivery Team

GPT KB + Git

Shared source of truth for decisions, evidence, and handoffs

Show specialist lanes

PM / Delivery

Specialist delivery planning

Architecture

System boundaries and dependencies

QA

Delivery failure modes and test readiness

Release

Release-readiness critique

Workflow / Implementation

Scoped delivery support

Data Team

Specialist Review Perspective

View capability detail

UX / UI / IA

Specialist Review Perspective

Sonnet-style / Opus-style

Local review unless receipted

Supernova

Business / Opportunity Intelligence

Execution Control Role Cards

These roles govern scope, authorization, validation, and claim safety. They are separate from Big Crew specialist delivery work.

Super Runner (Execution Control Layer)

Primary responsibility: Controls task boundary, dependency, authority, caveats, and stop conditions before work proceeds.

Evidence contribution: Shows whether dependency, authority, stop condition, unsafe scope, and caveat fields were declared.

Not allowed: Does not act as an autonomous platform, approve execution, claim production readiness, or make cost/replacement decisions.

Relationship to Big Crew: Governs the task boundary around Big Crew work; it is not a delivery specialist role.

Runner (Bounded Task Executor)

Primary responsibility: Executes approved bounded tasks without inventing new scope.

Evidence contribution: Shows files touched, scope adherence, validation-ready output, and handoff artifact status.

Not allowed: Does not add unapproved scope, self-approve output, or convert task completion into performance proof.

Relationship to Big Crew: Executes bounded work after scope is controlled; it is not the same as Big Crew implementation perspective.

Checker (Evidence & Claim Boundary Checker)

Primary responsibility: Validates evidence completeness, source-of-truth alignment, and claim boundaries.

Evidence contribution: Shows evidence gaps, claim drift, source-of-truth mismatches, validation status, and first-pass acceptance.

Not allowed: Does not replace Big Crew QA, approve production readiness, or claim benchmark, cost, speed, or public-proof outcomes.

Relationship to Big Crew: Checks whether claims are evidence-safe; Big Crew QA checks delivery quality and failure modes.

Supporting Policies

Detailed routing and budget policies are collapsed by default so the org-role hub stays focused on scan and navigation.

Role & Responsibility Policy
Role / teamResponsibilityOwnsDoes not ownBudget policyEscalation rule
Lyn — Owner / Decision AuthorityOwner review, risk appetite, and final approval for public, money, legal, privacy, reputation, health, career, and major architecture decisions.Final gate and protected decisionsRoutine execution or broad staffing before evidence existsSararin-gatedStop and ask when the decision is high-risk or hard to reverse.
Robert / GPT — Executive Orchestration LayerExecutive orchestration, synthesis, sequencing, decision framing, and source-of-truth alignment.Tradeoff framing, narrative coherence, and claim-boundary reviewOwner authorization, autonomous execution approval, or production-readiness claimsExecutive review laneEscalate to Lyn when the decision touches protected categories or publication authority.
Super Runner (Execution Control Layer)Controls task boundary, dependency, authority, caveats, and stop conditions before work proceeds.Scope, authorization check, dependency visibility, and stop-condition preservationAutonomous platform authority, production readiness, cost decisions, or replacement decisionsControl laneStop when owner authorization, unsafe scope, missing dependency, or claim boundary is unresolved.
Runner (Bounded Task Executor)Executes approved bounded tasks without inventing new scope.Scoped output, files changed, validation-ready handoff, and implementation caveatsScope creation, self-approval, benchmark proof, or performance claimsBounded execution laneReturn to Super Runner when scope changes or required authorization is missing.
Checker (Evidence & Claim Boundary Checker)Validates evidence completeness, source-of-truth alignment, and claim boundaries.Evidence gaps, source-of-truth mismatches, claim drift, and validation statusBig Crew QA replacement, public proof approval, production readiness, or cost superiority claimsEvidence safety laneBlock or downgrade when evidence is missing, claims drift, or source-of-truth alignment fails.
Big Crew (Specialist Delivery Team)Specialist delivery work across PM / Delivery, Architecture, QA, Release, Workflow, and Implementation.Role-specific critique, implementation support, QA findings, and release recommendationsExecution-control authority, final release approval, or Lyn-protected decisionsSenior synthesis onlyUse senior model review only for named decisions with clear expected value.
ResearcherBounded external evidence scans for market context, public claims, and practice checks.Cited evidence, safe-claim boundaries, and research briefsFinal decisions, market certainty, or unsupported uniqueness claimsCheap/web firstEscalate only when synthesis changes a public claim or decision gate.
SupernovaOpportunity and monetization intelligence through advisor-style memos and trigger-based analysis.Business option framing, kill arguments, and monetization questionsActive execution before a concrete trigger existsTrigger-basedActivate only when the business question is bounded and reviewable.
Investment TeamPlanned investment advisory lane after Sararin defines scope, criteria, and evidence requirements.TBD investment workflow once scopedLive investment action, recommendations, or decisions todayNot activeRemain parked until scope and review policy exist.
optimize-workerRouting and workflow execution support for benchmark traces, fallback paths, and thin-slice automation.Structured runs, trace shape, and repeatable validation supportFinal interpretation, provider expansion, or autonomous production routingCheap/local firstEscalate when automation would create false confidence without trace quality.
Budget Policy

Role first, model second. Budget is allocated like workforce salary: use the cheapest reliable route for routine work and reserve senior review for named decisions with a clear expected value.

Cheap/local first
Tier 1

Use for formatting, task packets, trace updates, markdown cleanup, extraction, checklists, and low-risk structured work.

Senior synthesis
Tier 2

Use only when a named review, critique, tradeoff, or decision-quality output needs senior judgment.

Executive escalation
Tier 3

Reserved for rare foundation decisions and Sararin-protected categories; requires an explicit gate.

Implementation lane
Tier 4

Codex is used for repo and UI implementation by role fit, with files, validation, commit, and caveats returned.

No senior delegation without a named decision, expected value, cost cap, and abort condition.

Task Routing Policy
Repo or UI patch
Codex

Use a bounded task packet, then return files changed, validation, commit hash, and caveats.

Architecture or release critique
Big Crew

Use role-specific review for a named decision, then Sararin gates the final call.

External evidence scan
Researcher

Gather cited evidence and safe-claim boundaries; do not make the final claim.

Opportunity analysis
Supernova

Activate only when there is a concrete business trigger and a reviewable output.

Trace, handoff, or validation summary
Hermes

Keep the work structured, public-safe where needed, and synced back to the KB.

Money, legal, privacy, reputation, health, career, or major architecture
GPT / Sararin

Stop for executive review; no delegated role owns the final decision.