The foundation of the autonomous enterprise
AI is entering sociotechnical organizations it doesn't understand. Humans have always scrambled in the same way. No technology has ever supplied a satisfactory capability to understand, define and realize the organization coherently end-to-end.
Poesis grounds humans and AI in a coherent end-to-end system that manages Definitions of any sociotechnical organization — its DNA — to efficiently and continuously define, govern and generate that organization.
The DNA of the organization constitutes a living digital twin — the foundation for accelerating THINK, BUILD and RUN adequately, ultimately leading to the autonomous enterprise.
Human-governed Definitions that people and AI can act on.
Pain points
This is the central tension of industrial GenAI adoption. GenAI accelerates the production of artifacts, but not the production of Definition and governance required to viably deliver these artifacts.
The scarce resource is now more than ever the THINK: the AI and human capacity to define and govern these Definitions for BUILD to generate from and RUN to maintain from.
Affordances
Poesis resolves the historical THINK pathology — for AI, people, and programmatic tools, to accelerate THINK, BUILD and RUN at the expected rate and safety.
Work from shared, typed definitions
Establish a common semantic basis for decisions across authors and tools by connecting typed IT inventory, retained definition versions, shared vocabulary and cited dependency analysis. Direct authoring and reviewed sourcing are alternative inputs; integrated authorized selection, interchange and answer quality need acceptance, and accepted intent is not observed live state.
Review selected obligations together
Bring obligations from selected frameworks and internal policies into a shared scope so adopters can review applicability, overlaps and gaps against the same IT subjects. Sourced vocabulary and templates support planned composition and accountable interpretation; selection establishes neither conflict freedom, legal completeness nor compliance enforcement.
Make review and activation boundaries explicit
Connect proposals to accountable review and activation through governance grammar, lifecycle controls and human decision gates. Transition controls exist, while authenticated revision-bound decisions need identity and policy integration; sealing closes collection rather than approving definitions, and research affects model revision only through separately accepted Change Proposals.
Review candidates from selected sources
Build a reviewable account of IT and delivery work by comparing and reconciling code-derived and delivery-derived evidence as retained candidates with provenance. End-to-end acquisition, review and draft registration are planned; confidence is not truth, source history is not deployed reality, and draft promotion remains separate from acceptance.
Evaluate identified inputs against supported Norms
Produce decision-ready compliance appraisal by combining pure verdicts over pinned rules and inputs with scoped IT measures, findings and corrective review. Planned evaluation preserves unknown and not-applicable states; running-system claims additionally need deployed identity, fresh observations and coverage, continuous appraisal needs scheduling, and results are not compliance certification.
Generate artifacts through declared workflows
Produce grounded artifacts from selected definitions through bounded workflow orchestration, declared operations, artifact contracts and accountable review, retaining outputs and execution evidence in a local delivery baseline. Local controls exist; integrated generation, effect safety and authenticated human gates need acceptance. Reviewed documents do not establish tested code, authorized deployment or operational feedback; hosting is optional.
Implemented for the stated scope; Partial where only part of it is; Planned for Poesis commitments still being implemented.
Usage
To the people and the machines who think, build, and run it. Definitions stop being prose: a structured causal context that can be reasoned over, taught, governed, simulated, and continuously evaluated.
Retrieve the decided revision, recorded identity, supporting evidence and conditions relevant to a historical decision, retaining attribution gaps.
Inspect dependent definition eligibility after a referenced foundation is invalidated and determine which continued uses require renewed review.
Agree retention purpose, duration, access and disposal constraints for decision evidence, distinguishing agreed policy from demonstrated enforcement.
Agree a versioned vocabulary and its constraints with adopters, using examples and counterexamples to expose incompatible interpretations.
Compare exported and imported definitions against an explicit conformance profile, including identity, lifecycle meaning and nonconformant draft preservation.
Review resource/action grants against a bounded assignment and its expiry, then check those grants and the target baseline before mediated work. Mandate review is planned; grant checking alone is not authenticated human approval.
Compare candidate model routes against workload-specific quality, cost and disclosure criteria using retained evaluation evidence.
Verify that a target host receives the pinned instructions and declared checks, distinguishing delivery from model adherence before migration.
Interpret the primitive, typing, governance-grammar and lifecycle clauses relevant to a proposed decision, identifying permitted representations and constraints without claiming that a processor enforces them.
Select sourced HTTP, IT-domain and provenance concepts for a service description, checking the meaning and attribution of each assertion. Vocabulary does not implement the service or prove its behavior.
Prepare a source-linked shortlist of architecture, quality and regulatory requirements for a bounded IT review.
Trace a non-normative modeling question to its explanations and recorded decision status.
Create, query and revise typed definitions through the REST contract, inspecting validation, revision and access responses without treating a successful draft write as approval.
Post a candidate to a declared panel and inspect schema, identity, confidence-envelope and rejection responses under the client/server contract. Acceptance into collection is not definition approval.
Request sealing or abandonment of a contribution collection and inspect concurrent-intake outcomes and snapshot readiness. Closing collection does not approve or activate a definition.
Retrieve stored candidates and their captured evidence to inspect provenance and conflicting assertions, making missing audit, integrity or promoted-draft links explicit.
Resolve eligible steps from workflow configuration and recorded outcomes on a supported adapter; leave dispatch to the host rather than treating resolution as execution.
Check a proposed write set against a matched schema and inspect structured-data or Markdown-metadata violations before writing. Passing validation does not establish prose quality or acceptance.
Read correlated session-journal records to reconstruct captured harness invocations and their outcomes, distinguishing absent records from evidence that no action occurred.
Select portfolio, program, team and specialist roles for a delivery responsibility and make their boundaries explicit without treating a role definition as an authenticated authority grant.
Identify the versioned instructions governing a role's engineering work and distinguish advisory guidance from requirements enforced by separate executable gates.
Collect independent candidates under declared panel contracts and revisit their retained provenance to compare conflicting assertions. Contribution acceptance and evidence retention do not themselves seal a board or promote an approved definition.
Represent a scoped governance concern using the fixed primitive roles and domain Archetypes together, exposing relationships and concerns the chosen representation cannot express. A valid representation does not establish universal model adequacy.
Define domain vocabulary over the fixed subject types, identify its versioned dependencies and specify how typed definitions pass through correction, review and lifecycle change. Processors implement the specified controls.
Resolve eligible work, supply configured guidance, select a model route and correlate captured invocations for a supported local execution. Host dispatch and probabilistic generation remain distinct from deterministic workflow controls.
No delivered use case names this actor type yet. Browse all use cases →
Bring relevant context and affected definitions into a shared review, making dependency paths and missing evidence visible before deciding.
Use planned scoped evaluation to surface findings earlier, with observation coverage, freshness and unknown states visible to reviewers.
Build toward delegated workflows with explicit obligations, decision boundaries and controlled effects, retaining evidence for intervention and recovery.
Reuse shared context and generated artifacts so effort can shift from repeated assembly to curation and review. Net savings depend on the workflow and compute costs.
Bring linked snapshots, decisions and provenance into a stated audit scope, with completeness assessed against that review rather than inferred from history alone.
Qualities
A governance layer earns trust only if you can leave it. The Definition model is an open, inspectable standard rather than a proprietary format; execution runs inside a control boundary you define, on infrastructure you already operate — cloud where it serves you, never as a condition. And because governance claims are worthless without their limits, each quality below states what it guarantees and where that guarantee stops.
Governance does not depend on one provider
Model profiles support selection for bounded tasks. Each host mapping and model change needs compatibility and quality checks.
Define the data boundary explicitly
Local execution supports control, but models, tools, telemetry, remotes and backups can transmit data. Egress and retention policies remain necessary.
Managed infrastructure without ceding control
Deploy services through managed infrastructure where it makes sense, while the control boundary stays explicit rather than assumed.
Governance you can inspect, not a black box
GSM is a candidate common Definition model. Interchange requirements support inspectability; compatible consumers and draft-preserving migration still need acceptance.
Trust cannot be proprietary
GSM specification content exists in source, with publication pending. SAF local tools use Apache-2.0; SIE and the planned bridge have separate licenses. Inspect each product and its terms.
Planned candidates from pinned revisions
Supported analysis is planned to produce reviewable candidates. Acquisition, conflict reduction and governed registration are distinct steps.
The interaction surface of your estate
Planned adapters describe declared interaction surfaces with explicit coverage. Contracts alone do not prove actual traffic or applicable obligations.
Planned dependency evidence
SBOM analysis identifies declared dependencies in a supported scope. Presence does not prove deployed use, reachability or compliance.
Separate intent from observations
Planned infrastructure sourcing reads declared configuration. Deployed identity, observation time and coverage are needed to describe what runs.
Planned The sourcing bench is specified and in build. Named-tool connectors follow as further knowledge sources feeding the same governed model.
Services
Evaluate Poesis on a real problem, shape your path toward enterprise autonomy, or prepare the workflows and context that make AI useful. Our service roadmap extends from transformation and context engineering to managed ITIP and SIE, and GSM practitioner certification. Start today with a Poesis Pilot by agreement; all other offers are planned.
Prove the value in your own context.
Evaluate Poesis against a real organizational problem, with an agreed scope, a baseline and evidence that supports your next investment decision.
Design the organization that can govern its own evolution.
Connect strategy, decision rights, organizational design and feedback into a practical path toward greater autonomy, without losing human accountability.
Find the work worth delegating. Make delegation accountable.
Discover and prioritize agentic use cases, redesign the workflows around them, and define how results, cost and risk will be evaluated.
Give humans and AI a shared understanding of the organization.
Turn fragmented organizational knowledge into structured, governed context: purposes, responsibilities, dependencies, rules and the evidence behind them.
Turn IT knowledge into better-governed decisions.
A planned managed IT Intelligence Platform for application and service context, dependency analysis, governance review and evidence-led IT decisions.
Build on governed definitions, not another isolated knowledge store.
A planned managed Systemic Intelligence Engine for builders integrating GSM definitions, lifecycle governance and controlled execution into their own applications.
Demonstrate the judgment behind a coherent organizational model.
A planned qualification for practitioners applying systemics, GSM modeling and governance through assessed cases and reviewable evidence.
Platform
Specification, source content, local controls and planned integrations contribute in different ways. Explore each product for feature scope and historical version metadata.
The cross-solution capstone: affordances realized by contributing capabilities across every solution, and the qualities the platform bears as a whole.
A planned IT workspace for shared definitions, evidence review, scoped appraisal and generation.
Bring IT definitions, evidence and decisions into one shared workspace.
Turn inspected code revisions into evidence-backed definition proposals.
Definition, lifecycle, execution and intake foundations, with scoped integration work remaining.
Keep system definitions versioned, accessible and governed through change.
Connect governed definitions to execution and planned Norm evaluation.
Bring evidence and competing proposals into a shared review space.
Candidate definition semantics, selected sourced vocabulary and non-normative research.
Define systems and governance with a shared, extensible language.
Start governance modeling from sourced vocabulary and reusable templates.
Explore and challenge the ideas shaping GSM.
Version convention Product versions shown are current, not historical milestones. Implemented and Planned describe feature, capability and affordance scope, not hosted availability or certification.
Poesis