KATLAS Operating Model · Roles, wallets and receipts

From synthetic roles to governed workflows.

See how KATLAS roles, wallets, evidence sources, authority boundaries and receipts turn an AI-enabled idea into a deployable operating model.

Starting point

Lab Pathway does not begin by connecting to live systems or ingesting private data.

It begins by designing the operating model: who participates, what evidence exists, who controls it, what AI may suggest, who may approve action, which systems may be touched, and what receipt proves the outcome.

Section 01 · Wallets as roles

A wallet represents a role in the workflow.

A KATLAS wallet can represent a person, team, organisation, device, system, AI agent or external partner.

Each wallet can carry custody, authority and receipt responsibilities.

The wallet does not require raw private data to be centralised. It governs what can be requested, shared, withheld, approved or evidenced.

The role controls the boundary. The receipt proves the action.

Section 02 · Synthetic first, live later

Start synthetically. Deploy only when approved.

The playground designs the roles. KATLAS operationalises them once governance, evidence boundaries and authority gates are approved.

Synthetic playground

Designed, not deployed

  • Representative roles
  • Fictional or synthetic data
  • Simulated requests
  • Simulated approvals
  • Simulated receipts
  • No live system access

Live KATLAS R2R workflow

Operationalised once approved

  • Live role-to-role identities
  • Explicit permissions
  • Approved evidence rights
  • Governed requests
  • Authority gates
  • KATLAS receipts
  • Deployment controls

The playground designs the roles. KATLAS operationalises them.

Section 03 · Source registration

How existing systems plug in.

Legacy and new systems are first registered as evidence sources. This does not require data ingestion.

Evidence remains at source. KATLAS governs what can be requested, summarised, withheld or approved.

Each source is described by

  • /01source type
  • /02owner
  • /03sensitivity
  • /04permitted use
  • /05AI-use boundary
  • /06authority gate
  • /07receipt requirement
  • /08integration status

Example sources

CRMERPdata warehousedocument storeGit repositoryAPIspreadsheetIoT or wearable deviceLLM or SLMdashboardexternal partner platform

Live integrations are configured only where explicitly approved. The playground does not require API keys, credentials or production access.

Section 04 · CAR matrix

Turn the workflow into CAR.

Each role's row makes its custody, authority and receipt responsibilities explicit. The matrix below shows three representative roles.

RoleCustodyAuthorityReceipts
SponsorSponsor brief and budget boundaryApprove lab scope and next-stage funding routeSponsor pathway brief · approval receipt
API ownerAPI documentation and access rulesApprove or refuse connectionIntegration approval · refusal receipt
AI assistantNoneAdvisory onlyModel route · recommendation boundary

Section 05 · Skills and capabilities

What capabilities are needed?

The operating model is delivered through a small mix of business, technical, governance and academic roles. Each row maps a capability to the kind of contributor who typically owns it.

Sponsor objective
COO · innovation lead · department sponsor
Workflow knowledge
Business owner · operational team · domain expert
Data / source ownership
System owner · data steward · IT
Authority mapping
Governance · legal · compliance · risk
AI-use boundary
AI lead · risk lead · domain expert · KATLAS
Integration design
API lead · architect · DevOps · system owner
Wallet / R2R design
KATLAS
Education and adoption
Education / adoption partner, where appointed
Validation and whitepaper
Academic contributors, where appointed
Deployment control
DevOps · security · KATLAS · client technology team

Section 06 · Tools and artefacts

What the operating model produces.

A small set of buyer-friendly artefacts that capture roles, sources, boundaries, routing and receipts — ready to take into a budget, security or deployment review.

  • /01Role map
  • /02Source register
  • /03CAR matrix
  • /04AI-use boundary
  • /05Model-routing view
  • /06Authority gate map
  • /07API / integration map
  • /08Receipt template
  • /09Deployment runbook
  • /10Validation or whitepaper outline
  • /11KATLAS receipt viewer

Section 07 · Technical view

For technical teams.

The operating model can be expressed through structured configuration and APIs: role definitions, source records, permissions, request states, approval events and receipt payloads.

This makes the governance model explainable to product owners, architects, API developers, DevOps teams and security reviewers.

Pseudo objects · no real secrets · illustrative only

Role object

{
  "role": "API owner",
  "custody": ["API documentation", "access policy"],
  "authority": ["approve_connection", "refuse_connection"],
  "receipts": ["integration_approval", "integration_refusal"]
}

Source object

{
  "source": "CRM",
  "owner_role": "Customer data owner",
  "sensitivity": "confidential",
  "permitted_use": "summary only",
  "ai_boundary": "no raw customer records",
  "authority_gate": "data_owner_approval_required"
}

Receipt object

{
  "event": "integration_request_reviewed",
  "actor_role": "API owner",
  "decision": "approved_for_synthetic_pilot",
  "evidence_shared": ["API schema summary"],
  "evidence_withheld": ["production credentials", "customer data"],
  "verification_state": "simulated"
}

Illustrative pseudo-objects · API keys, credentials and live integrations are not required for the playground · Live KATLAS R2R workflows are configured only once governance, evidence boundaries and authority gates are explicitly approved.

Continue

See the platform thesis and the example walkthroughs.