Understand complex work
Combine documents, conversations, assets, rules and workflows into a coherent operational picture.
AIOS / SAM · global architecture · active development
Not a chatbot.
Global operating intelligence.
SAM converts documents, conversations, assets, rules and workflows into governed understanding and prepared action. She is designed as a domain-extensible system: able to be built for any field whose knowledge, processes, evidence and authority can be responsibly defined and commissioned.
Self-awareObserves capability, limitation and stateSelf-evolvingProposes evidence-led improvementHuman-governedAuthority remains accountable
Capability map
What SAM is designed to do.
Reason with governance
Test evidence, uncertainty, responsibility and authority before recommending or preparing action.
Interpret obligations
Map standards, contracts and regulatory requirements to assets, work, evidence and accountable people.
Prepare controlled action
Build plans, workpacks, reports and next steps for authorised execution through governed boundaries.
Build audit-ready evidence
Preserve provenance, decisions, approvals, results, confidence and rollback information in inspectable receipts.
Observe her own state
Maintain a model of capabilities, limitations, dependencies, drift, uncertainty and current operating health.
Propose her own improvement
Generate questions and bounded learning proposals from failures, gaps, contradictions and repeated user friction.
Work across tools
Route approved models, services and tools through identity, permission, verification and audit controls.
Run locally
Support local or controlled hybrid operation where privacy, resilience, sovereignty or customer independence matter.
Reveal the interface only when needed
Let people ask for outcomes naturally, then expose conventional screens for review, confirmation or expert control.
Find risk and opportunity
Prepare source-grounded intelligence on compliance debt, anomalies, tenders, leads, partners and emerging change.
Remain honestly bounded
Say ‘I don’t know’, preserve unresolved questions and refuse to represent an unverified state as fact.
Built for any definable domain
Global foundation.
Domain-specific intelligence.
Domain pack
Define the terminology, entities, documents, workflows, rules, evidence models and decision boundaries of the field.
Jurisdiction overlay
Apply the relevant local standards, regulation, language, practice and authority—without assuming one market transfers to another.
Specialist stewardship
Domain experts shape, challenge and approve the intelligence required for responsible real-world operation.
Commissioning proof
Test the bounded capability, gather acceptance evidence and commission it for its intended environment.
SAM does not claim instant expertise in everything. Her architecture can extend into any definable domain, then earn competence one governed capability at a time.
Truth before theatre
Every capability has a state.
DesignedArchitecture and controls defined→BuiltImplemented with test evidence→DeployedMaterialised in an environment→CommissionedAccepted through live proof
AIOS/SAM is a developing system containing capabilities at different states. SprintLab does not treat code, configuration or a demonstration as proof of complete operational commissioning.
The intended experience
Ask for the outcome.
Keep control of the decision.
A contractor should be able to ask, “What compliance work is overdue, why, and what evidence is missing?”
An investor should be able to ask, “Which claims are commissioned, which remain ambitions, and what evidence supports each answer?”
An AI company should be able to ask, “Can our model operate here, under what authority, and how will its behaviour be verified?”