Capability axis
Identity, state, valuation, propagation, forensics, decision, evidence and authoring.
Governed analytical architecture
Worlds, capabilities, surfaces and applications—over one disciplined substrate.
Kindynos separates domain content from reusable analytical machinery while preserving model identity, state, evidence, stakeholder consequence and refusal.
Canonical view: EVA as the broad operator environment; authored worlds; specialist intelligence; application capabilities; and EOS/EPL as a modular execution and authoring architecture within the Kindynos technology estate.
World axis
Assets, stakeholders, events, contracts, conditions, regulations and model bindings.
Surface axis
Graphical environment, API, chat, MCP, CLI, workbench and reports.
Application axis
EVA, KAILA, CSUITE, STAR and specialist domain intelligence.
The platform thesis
Reusable capabilities gain domain meaning through authored worlds.
Kindynos organizes the estate as a capability-by-world matrix. Shared analytical machinery combines with each domain’s authored content, resolved bindings, exercised paths and evidence.
Evidence-based technology assessment — this view synthesizes reviewed code, documented architecture and available execution evidence to show the breadth and relative implementation depth of Kindynos capabilities across representative domains. Deployment assurance is assessed separately for each implementation and operating environment.
Why this mattersEach column shows how reusable capabilities combine with domain-specific assets, stakeholders, events, contracts, models and evidence. This makes the work required to deepen a world visible and actionable.
The governed substrate
What must remain true beneath every useful answer.
The architecture keeps authoritative computation inspectable and governed, while language models assist authoring, retrieval and explanation.
Canonical identity
Assets, parties, models, positions and worlds remain identifiable across time and systems.
State & path dependence
Earlier events and decisions can write state that later events read.
Model & input precedence
Applicable versions, required inputs and source hierarchy are explicit.
Difference valuation
The same position is valued before and after the event; the consequence is the difference.
Per-position dispatch
Each position can resolve its own model and produce a stakeholder-specific result.
Provenance & refusal
Evidence accompanies the result; missing required machinery produces a named refusal.
Current implementation foundation
EVA is the integrated operator environment.
EVA is the broadest visible Kindynos environment. It combines registry and browse layers, event-impact analysis, ecosystem authoring, model inputs, graph capabilities, investment tooling, forensics, decision analysis and evidence surfaces.
Its interfaces organize a broad capability estate into registries, guided workflows, analytical workbenches and specialist views.
- Event Mode: apply events to a represented world
- Persona Mode: take the viewpoint of a stakeholder
- Workbench: assemble and inspect analytical workflows
- Multiple surfaces: GUI, API, chat, MCP and reports
One event, four lenses
Risk, forensics, investment and decision share the same represented world.
This taxonomy maps directly to the product architecture.
World authoring
New domains become governed worlds within a shared analytical architecture.
Each world defines its own assets, stakeholders, events, positions, contracts, conditions, regulations and models. Kindynos binds those domain structures to reusable capabilities for identity, valuation, propagation, forensics, investment analysis, decision support and evidence.
Domain-specific models, validated data and evidence deepen each world while the underlying analytical machinery remains shared.
EOS / EPL
A modular execution and authoring architecture.
EOS is a domain-independent kernel. EPL is a declarative authoring language compiled into an intermediate representation for governed execution.
EOS/EPL complements EVA by providing declarative world authoring, verification, compiled intermediate representation, governed state transitions, model boundaries, effects and ledger records. Its modular design supports selective interoperability, shared components and additional execution surfaces within the Kindynos estate.
AI boundary
Language models assist; governed execution remains authoritative.
Generative systems support authoring, retrieval, classification, explanation and narrative. Authoritative valuation, state commitment, model-version choice and stakeholder consequence remain within governed analytical paths.
Assist
Authoring, discovery, interface interaction, explanation and report preparation.
Verify
Deterministic or reproducible execution paths evaluate the authoritative analytical result.
Refuse
Missing models, inputs or bindings remain visible rather than being converted into plausible fiction.
Apply the architecture to a consequential problem.
A focused engagement can begin with one event, portfolio, case, decision or domain world and the systems it crosses.