Intuition Machine
Machines for the self-learning enterprise
A family of specialised intelligence systems that help an enterprise understand itself, coordinate its work, and improve continuously through experience.
DiscoverConnectUnderstandJudgeActLearn
Learn the enterprise by doing useful work in it
Traditional enterprise software assumes an organisation can describe its structures, processes, policies and requirements explicitly before automation begins.
Real enterprises are different. Their knowledge is scattered across systems, documents, people and institutional habit. Processes go undocumented, rules sometimes conflict, and the organisation keeps changing underneath both.
So instead of asking an enterprise to model itself before it can use AI, our machines discover, validate and formalise organisational knowledge progressively, through useful work. As that knowledge accumulates, the enterprise becomes steadily more able to understand its own operations and automate them reliably.
The work the enterprise needed done, done now.
Validated knowledge the next operation starts from.
Eight machines, one foundation
Each machine specialises in one aspect of enterprise intelligence and answers one fundamental question. Together they form a system that can progressively understand, operate and improve an organisation.
Construct and maintain reliable organisational knowledge
Turn knowledge into justified decisions, execution and improvement
Deliver coordinated outcomes across people and agents
Sense Machine
Make the invisible enterprise visible.
What exists, and what is happening?
Sense Machine discovers how an enterprise actually operates. It examines existing systems without first requiring the organisation to produce a comprehensive ontology or formal architectural model.
Using connectors, crawlers, static analysis, process mining and AI-assisted interpretation, it reconstructs the relationships among applications, data, business processes, organisational responsibilities and operating rules — both the structures that are documented and the implicit relationships that emerge from operational evidence.
Principal capabilities
- Enterprise crawling and system inventory
- Software and data architecture reverse engineering
- Process and workflow discovery
- Relationship and dependency extraction
- Identification of implicit rules and practices
- Detection of undocumented assumptions and contradictions
- Targeted elicitation of tacit knowledge from domain experts
- Produces
- Evidence-backed observations, candidate semantic assertions, dependency maps and explicit knowledge gaps.
- Creates value by
- Reducing the cost of understanding complex, poorly documented enterprise environments.
Sense Machine is the primary way enterprise intelligence is bootstrapped from messy organisational reality.
Connection Machine
Connect organisational claims to reality.
How are claims connected to evidence?
Connection Machine establishes the evidence underneath enterprise knowledge. It maintains witness graphs that connect claims, observations, interpretations and decisions to what supports them.
It shows not merely that two things are related, but why that relationship is believed and where the supporting evidence comes from. That matters most when AI-generated interpretations circulate through several systems and begin to look independently corroborated.
Principal capabilities
- Evidence and provenance tracing
- Witness graph construction
- Claim-to-source grounding
- Dependency and derivation tracking
- Detection of unsupported claims
- Identification of circular or duplicated evidence
- Traceability across organisational decisions
- Produces
- Witness graphs, provenance chains, grounding assessments and evidence-dependency structures.
- Creates value by
- Making enterprise intelligence traceable, auditable and resistant to unsupported or self-reinforcing conclusions.
Connection Machine provides the grounding that trustworthy shared knowledge requires.
Context Machine
Give every decision the context it deserves.
What do we know, and what matters here?
Context Machine maintains an evolving representation of what an enterprise knows, believes, requires and intends, preserving organisational meaning across systems, teams and time.
Unlike a conventional knowledge repository, it represents uncertainty, conflicting interpretations, authority, applicability and change explicitly. It keeps three kinds of knowledge distinct rather than forcing agreement between perspectives:
- Descriptive
- What exists or happens.
- Normative
- What is required or permitted.
- Intentional
- What the organisation is trying to accomplish.
It also assembles task-specific views, giving each agent the knowledge relevant to its work without exposing it to the full complexity of the enterprise.
Principal capabilities
- Persistent enterprise semantic memory
- Entity and relationship management
- Representation of assertions, conflicts and unknowns
- Context-sensitive interpretation
- Organisational authority and policy context
- Temporal knowledge and semantic versioning
- Permission-aware contextual retrieval
- Task-specific context assembly
- Produces
- Enterprise semantic models, contextual knowledge views, scoped assertions and decision-relevant context.
- Creates value by
- Making organisational understanding reusable across applications, teams and autonomous agents.
Context Machine preserves what the enterprise learns.
Judgment Machine
Determine what deserves to be trusted.
What conclusions are justified?
Judgment Machine establishes whether interpretations, conclusions and proposed actions are sufficiently justified. It distinguishes evidence from inference, confidence from correctness, and observed behaviour from legitimate authority.
Different situations need different kinds of validation. Some assertions can be checked deterministically; others need specialised classifiers, formal policies, empirical evaluations or an authorised human. Judgment Machine coordinates them.
Principal capabilities
- Evidence sufficiency evaluation
- Semantic consistency checking
- Policy and authority validation
- Conflict and contradiction analysis
- Task-specific eval generation
- Risk-sensitive decision thresholds
- Uncertainty assessment and abstention
- Human approval and escalation
- Specialised typed judgments
- Produces
- Validated assertions, judgment results, evaluation evidence, authorisation requirements and decision constraints.
- Creates value by
- Making enterprise AI decisions verifiable, accountable and governable.
Judgment Machine makes sure that accumulating information does not automatically become accumulating authority.
Intuition Machine
Know what kind of intelligence is needed.
How should this situation be handled?
Intuition Machine decides how each situation should be handled. Not every enterprise event needs the same cognitive process: routine transactions can run on deterministic rules, familiar exceptions need narrow judgments, unfamiliar situations call for investigation, and consequential decisions need human deliberation.
It dispatches work to the right mechanism, treating different kinds of cognition as distinct operational capabilities rather than routing everything through one general-purpose model.
Principal capabilities
- Situation classification and typed dispatch
- Selection of appropriate cognitive workflows
- Deterministic versus judgment-based routing
- Escalation and exception handling
- Selection among available processing capabilities
- Prevention of inappropriate routine processing
- Routing traceability and diagnostics
- Produces
- Typed dispatch decisions, selected cognitive workflows and escalation instructions.
- Creates value by
- Improving the efficiency and reliability of enterprise intelligence through the right cognitive specialisation.
Intuition Machine makes sure the enterprise applies the right kind of intelligence to each situation.
Agency Machine
Turn justified understanding into authorised action.
What are we authorised to do?
Agency Machine lets AI agents perform enterprise work within explicit organisational constraints. Rather than leaving each agent to interpret policy for itself, it compiles the applicable requirements, permissions and decision contracts into constrained execution environments — a semantic compiler and an enterprise meta-harness.
It separates the ability to perform an action from the authority to perform it.
Principal capabilities
- Semantic compilation of requirements and policies
- Generation of specialised agent harnesses
- Scoped tool permissions and credentials
- Model and judgment-service integration
- Workflow orchestration and durable execution
- Human approval gates
- Runtime precondition verification
- Exception handling, recovery and rollback
- Execution tracing and auditing
- Produces
- Compiled decision contracts, constrained agents, governed workflows and action records.
- Creates value by
- Enabling reliable enterprise automation without giving up organisational control.
Agency Machine turns knowledge into safely delegated operational capability.
Learning Machine
Make experience improve the enterprise.
What should we learn or change?
Learning Machine closes the loop between organisational knowledge, actions and outcomes. It watches for changes in processes, systems, policies, decisions and results, and identifies when assumptions should be reconsidered, when new knowledge should be acquired, and when deployed workflows must be revalidated.
It does not simply accumulate information. It manages the controlled evolution of organisational understanding.
Principal capabilities
- Semantic drift detection
- Operational feedback analysis
- Detection of obsolete or conflicting knowledge
- Initiation of semantic learning transactions
- Dependency-aware impact analysis
- Validation of proposed semantic revisions
- Regression evaluation of affected agents
- Knowledge refinement and retirement
- Measurement of semantic reuse and economic benefit
- Produces
- Semantic revisions, learning transactions, change-impact reports, updated validations and improvement recommendations.
- Creates value by
- Keeping enterprise intelligence from going stale while turning operational experience into reusable capability.
Learning Machine makes the enterprise’s intelligence compound rather than decay.
Coreography Machine
Enable people and AI agents to act as one coherent organisation.
How do we act coherently together?
Coreography Machine coordinates complex work across human teams, software systems and AI agents. It goes beyond conventional orchestration, where a central controller directs every participant.
Instead it relies on shared context, commitments, protocols and events that let distributed participants coordinate without a single controlling intelligence. People and agents work in shared project and organisational contexts — observing progress, making commitments, negotiating dependencies and responding to change — while each keeps their own responsibilities and delegated authority. It brings together agentic technical programme management and multiplayer AI.
Principal capabilities
- Multiplayer human–agent workspaces
- Autonomous technical programme management
- Cross-functional dependency discovery
- Commitment and decision tracking
- Distributed, event-driven workflow coordination
- Detection of conflicting plans and responsibilities
- Human–agent negotiation and escalation
- Shared operational state and progress monitoring
- Continuous capture of coordination knowledge
- Produces
- Coordinated activities, shared commitments, resolved dependencies, decision histories and reusable organisational knowledge.
- Creates value by
- Reducing organisational friction, coordination overhead, delivery delays and fragmented decision-making.
Coreography Machine is where the whole architecture meets the organisation: it lets the enterprise act collectively while getting steadily better at understanding and coordinating itself.
One foundation. Many forms of intelligence.
Common Ground is the architecture all eight machines share. It is not another agent, application or enterprise database. It is the common system through which knowledge, evidence, authority, uncertainty, decisions and learning are represented consistently — so no machine has to reconstruct or reinterpret organisational meaning on its own.
Entities, assertions, relationships, contexts, policies, decisions, commitments, outcomes and explicit unknowns.
How observations, claims, interpretations and decisions connect to where they came from.
When knowledge can be trusted, who may authorise a change, and where an interpretation applies.
The history of organisational meaning: definitions, obligations, ownership and dependencies as they change.
One standard way to investigate uncertainty, resolve contradictions, validate new understanding and publish it.
The interfaces through which the machines discover, interpret, validate, consume and revise knowledge.
Shared context, not one interpretation
Departments may legitimately use different definitions, priorities and operating rules. Common Ground represents those differences consistently, records the context each belongs to, and identifies when reconciliation is actually needed.
Its purpose is not to manufacture consensus. It is to make agreement, disagreement, uncertainty and authority explicit — and actionable.
How the enterprise learns, one question at a time
The essential operation on Common Ground. A transaction opens whenever work reveals an uncertainty, a contradiction, a missing dependency or an obsolete assumption.
- Define the unresolved semantic question.
- Collect the relevant evidence.
- Develop and compare interpretations.
- Identify the organisational authority that applies.
- Validate, reject, or preserve the uncertainty.
- Record the resulting semantic change.
- Determine which activities and agents depend on it.
- Sense Machine can open one through discovery.
- Connection Machine supplies the evidence.
- Context Machine maintains the knowledge involved.
- Judgment Machine evaluates proposed resolutions.
- Intuition Machine routes unfamiliar situations to the right process.
- Agency Machine applies validated decisions through controlled action.
- Learning Machine manages the resulting change in meaning.
- Coreography Machine creates new occasions for learning through coordinated work.
Common Ground is what makes these capabilities one intelligence system rather than eight products.
An interconnected system, not a pipeline
Sense investigates how the organisation really works; Connection grounds what it finds in evidence.
Context integrates knowledge, keeps competing interpretations, and builds the view each task needs.
Judgment tests whether conclusions are justified; Intuition picks the right process for each situation.
Agency executes within constraints; Coreography coordinates people and agents.
Learning turns new evidence, corrections and outcomes into validated knowledge on Common Ground.
More useful work → more validated understanding → better decisions → more capable automation → more useful work
The cycle only pays if evidence quality, semantic integrity, legitimate authority and economic efficiency hold all the way round.
You do not need all eight to get value
Each machine is a distinct capability. Start where the problem is; as deployments widen, Common Ground becomes more valuable, because knowledge validated once is reused by every new workflow.
An evidence-grounded technical programme manager
Our first integrated application combines Sense, Connection, Context and Coreography: a programme manager that discovers dependencies, exposes conflicting requirements, coordinates their resolution, and preserves what the organisation learns along the way. Judgment supplies the validation controls and Intuition the routing; Agency and Learning then widen its autonomy and adaptation.
The principle is the same one behind our first governed workflow offer: prove the economic benefit of accumulating reliable organisational knowledge in one place before attempting an enterprise-wide transformation.
Agents are not what enterprises lack
Enterprise AI has concentrated on better models, giving agents tools, and increasingly elaborate workflows. These help, but they are not enough.
What is missing is a continuously maintained, machine-usable representation of how the organisation actually operates — its obligations, authority, goals and accumulated experience. Without it, agents rediscover context again and again, reproduce contradictory assumptions, decide under outdated policies, and struggle to coordinate across organisational boundaries.
The machines provide the specialised capabilities. Common Ground keeps their shared understanding. Coreography Machine makes that intelligence useful across the organisation, and Learning Machine makes sure experience produces improvement rather than more semantic debt.
From fragmented knowledge to shared intelligence
- Sense Machine discovers reality.
- Connection Machine grounds understanding.
- Context Machine preserves meaning.
- Judgment Machine establishes validity.
- Intuition Machine chooses the right cognitive process.
- Agency Machine enables authorised action.
- Learning Machine turns experience into improvement.
- Coreography Machine enables collective intelligence.
- Common Ground connects them into one coherent system.
From isolated agents to coordinated action. From automation to continuous organisational learning.
Two machines, grounded in the kernel
Connection Machine and Intuition Machine come directly out of our research on how organisations learn. Here is the formal idea behind each.
What a witness graph is for
A convention-level claim is grounded exactly when a descent chain with provenance connects it to encounter-level records: the claim links to an interpretation, which links to a log, which links to something that actually happened. No such chain means no witness, and no witness means capture-vulnerable.
Connection Machine instruments that chain. It ingests the trace store, indexes the four strata of witness depth, and reports — per claim — whether a descent path to evidence exists and how many hops long it is. Because the test is mechanical, anyone can run it on any assertion.
What it cannot do is eliminate the remainder. Convention is never fully groundable, so a narratable margin is permanent. The job is to make the groundable portion commonly computable, and to say plainly which claims fall outside it.
Why dispatch needs a guard
A strange complaint arrives on Monday morning. It might be an operations matter, an audit trigger, or early market intelligence. Every loop is healthy; the diagnostic layers examine each one, find nothing, and return nothing — because the failure was that the item reached the wrong loop.
Intuition Machine makes the selector explicit. Every route is typed by level and column. Rules are marked as compiled or live, and the mix is matched to stage — compiled selection is appropriate standardisation at commodity stage and lethal at genesis.
Two guards are structural rather than configurable. Routes that must read quality-level content cannot be fully compiled, so no rulebook becomes the sole gate on quality-sensing. And every compiled rule set is itself a standing convention, so it carries a witness and an owner — because orphaned routing rules are the dead hand at the front desk.
The kernel is verified — seven axioms, nine load-bearing theorems, 43 of 43 checks passing. The selector construct is newer and labelled FRAMEWORK, with its formalisation still an open problem.
Start with one workflow.
We map your constraints, build a governed AI workflow alongside your existing operations, and prove the results before autonomy expands.