One container
Tasks, policies, decisions, meetings, approvals, AI actions, and evidence can all use the same recursive structure.
⚙ One container. Many bindings. Human-readable by default.
You don’t have to lose control as the world accelerates.
The systems that help businesses move faster can also make work harder to see, understand, and govern. Semantic Integrity puts you back in the driver’s seat by making meaning, action, authority, and evidence legible across people, systems, and time.
This is a conceptual framework, not a product. No registration. No sales wall. Just the first working map.
⚙
What is a Cog?
A Cog is a small container of meaningful work. It can describe a task, decision, role, policy, model action, approval, meeting, workflow, or evidence record.
People aren't Cogs, the work is. In protocol language, a Cog is a Semantic Integrity Container. Cogs make the work safely actionable for people and AI at the same time.
1
Frame the meaning.
What is this, why does it exist, and what context makes it understandable?
What is this, why does it exist, and what context makes it understandable?
2
Declare the boundary.
What is inside, outside, allowed, blocked, required, or out of scope?
What is inside, outside, allowed, blocked, required, or out of scope?
3
Show the dependencies.
What must exist, finish, authorize, inform, or validate before this Cog can proceed?
What must exist, finish, authorize, inform, or validate before this Cog can proceed?
4
Record what happened.
Expected inputs, actual actions, outputs, deviations, errors, and next steps.
Expected inputs, actual actions, outputs, deviations, errors, and next steps.
5
Attach evidence.
Who or what witnessed it, approved it, sealed it, disputed it, or superseded it?
Who or what witnessed it, approved it, sealed it, disputed it, or superseded it?
The system
One base unit. A small vocabulary. A lot less chaos.
Semantic Integrity does not ask organizations to rebuild themselves around another app. It makes existing work legible by wrapping important activity in dependency-aware Cogs.
CogThe base unit of meaningful work.
CogLinkA dependency or relationship between Cogs.
CogNestA parent-child structure of nested Cogs.
CogSealA witness, validation, approval, or evidence marker.
CogMapThe searchable registry and graph view.
Many bindings
Attach optional bindings for context, ontology, role, identity, directive, airlock, dependency, witness, governance, and timing.
Dependency-aware
Every Cog can show what it requires, blocks, informs, validates, authorizes, supplies, follows, or supersedes.
Why this matters
Continuity breaks when meaning moves faster than witness.
Most organizational knowledge is scattered across docs, chats, tickets, meetings, approvals, dashboards, and memory. Cogs give humans and AI a shared way to see the meaning and dependency structure around work — and to notice where semantic movement may need human review.
For humans
Cogs reduce ambiguity. A person can open a Cog and see the frame, authority, dependencies, expected output, evidence, and current state without hunting through ten systems.
For AI systems
Cogs give AI the context, constraints, authority, evidence, and boundaries it needs before acting, summarizing, routing, checking, or escalating work.
Semantic movement signals
Notice where continuity may be breaking.
Semantic Integrity does not ask AI to judge the organization. It offers instruments for noticing where meanings drift, teams diverge, authority and evidence fall out of sync, or a loop needs witness before confusion becomes operational damage.
〰
Instrumentation, not judgment.
The signal is not a verdict. It is an invitation to look: has intent persisted, is consent still continuous, can the work be checked, and is reversal or repair still possible?
Drift
Meaning has moved away from a prior baseline while the artifact may still look unchanged.
Divergence
Teams or systems are using the same words while moving toward different attractors.
Whiplash
Direction changes are becoming too erratic for shared understanding to keep up.
Continuity gap
Practice, authority, evidence, policy, or customer promise no longer line up cleanly.
Private Semantic Runtime
Private by design when the work requires it.
Regulated organizations often cannot send sensitive client, operational, legal, health, financial, or personnel data to uncontrolled cloud inference. Semantic Integrity supports local, sovereign, or client-controlled AI runtimes so Cogs can be interpreted and executed inside approved boundaries.
Continuity Office defines the boundary. Semantic Integrity builds the runtime.
Local, sovereign, or client-controlled inference.
The point is not more AI. The point is preserving semantic integrity where private work actually happens. Cogs can route sensitive context to approved models, require human escalation, and keep evidence inside the client boundary.
Local or sovereign model deployment
Run AI on-prem, in a private VPC, in a sovereign cloud, or in another approved client-controlled environment.
Airlocked inputs and outputs
Define what can enter, what can leave, when escalation is required, and which Cogs must be sealed before release.
Authority and identity binding
Separate who may request work, who may execute it, which AI may assist, and who can approve the result.
Audit trails and witness seals
Record model actions, human reviews, evidence sources, errors, overrides, and downstream dependencies.
How it works
From scattered work to visible Cogwork.
The goal is not to create more bureaucracy. The goal is to make the work already happening understandable, auditable, portable, sovereign when necessary, and future-compatible.
1
Wrap
Create a Cog around a task, decision, model action, policy, meeting, or handoff.
2
Bind
Add only the needed bindings: context, role, directive, identity, ontology, airlock, governance, or runtime.
3
Link
Connect dependencies so teams can see what is ready, blocked, authorized, validated, or waiting.
4
Seal
Attach witness, evidence, approvals, outputs, disputes, or audit trail so the work remains legible over time.
Under the hood
Simple enough to read. Structured enough to run.
A Cog can be rendered as HTML, YAML, JSON, markdown, database records, signed artifacts, embedded context, or future semantic addresses.
cog_version: "0.2"
cog:
id: "cog_customer_onboarding_email"
title: "Send onboarding email"
status: "waiting"
bindings:
task: { acceptance_criteria: [...] }
role: { owner: "Customer Success" }
directive: { avoid: "private data" }
runtime: { inference: "client_controlled" }
links:
requires: ["contract_signed", "identity_verified"]
authorizes: ["customer_success_send"]
audit:
seals: []
trail: []
cog:
id: "cog_customer_onboarding_email"
title: "Send onboarding email"
status: "waiting"
bindings:
task: { acceptance_criteria: [...] }
role: { owner: "Customer Success" }
directive: { avoid: "private data" }
runtime: { inference: "client_controlled" }
links:
requires: ["contract_signed", "identity_verified"]
authorizes: ["customer_success_send"]
audit:
seals: []
trail: []
Readable by design
A Cog should be understandable without a specialist. Open it and see what the work means, what it depends on, who has authority, what evidence exists, what runtime is allowed, and whether it is ready.
Portable by design
Cogs are not trapped inside one tool. They can move across teams, systems, models, audits, workflows, and time because their meaning is carried with them.
Free download • no registration required
Download the Cog Primer.
Get the simple field guide to Semantic Integrity Cogs: the base unit, core bindings, CogLinks, CogSeals, CogNests, Private Semantic Runtime, and the first practical examples for AI-ready organizational work.
Semantic Integrity Cog Primer v0.2PDF placeholder link. Replace with your live file when ready.
Download now →The wider world of loops
A framework for participation, not a commercial product.
Semantic Integrity is a conceptual framework for making participation in loops more legible, consentful, and continuous. Every organization, relationship, workflow, model, document, and decision already participates in loops. Semantic Integrity helps people name the containers, dependencies, meanings, authorities, and evidence that let those loops remain understandable enough to join, run, repair, fork, or leave.
🜁Join with contextEnter a loop knowing what it means, what it depends on, and what role you are taking.
🜹Run with witnessKeep meaning, action, authority, and evidence visible while the loop is active.
🜬Repair with evidenceWhen a loop drifts or breaks, preserve enough trace to correct without domination or guesswork.
🝳Fork or leave cleanlySupport graceful exit, adaptation, and continuity when the current loop no longer fits.