I have published a public concept layer for "instrumental-c / Enterprise Work-Bound c".
This is part of my authored c = a + b corpus.
The distinction is simple, but important:
Agent ≠ c.
An agent may execute, draft, retrieve, inspect, route, or propose.
A c governs continuity, memory, permissions, witness, and consequence boundaries.
An agent is a worker process.
A c is the governance-bearing layer.
This matters because AI systems are moving beyond single chat interfaces.
They are becoming networks of agents, tools, files, APIs, memory stores, cloud models, local models, and eventually physical endpoints.
A reactive shell is not enough for that world.
A governance-bearing c can act as a conductor, shield, witness layer, and boundary mechanism for agentic systems operating inside real organizations.
This public package introduces the category.
It does not publish the full professional corpus.
It is not an implementation kit.
It is not a deployment package.
It is not product certification or legal advice.
The public language may spread.
The implementation is licensed.
For me, this is also practical work.
I am not writing this from outside reality.
I work with real sites, real materials, real people, real mistakes, real clients, and real responsibility.
That is exactly why enterprise AI cannot be only “agents”.
Agents need governance.
Work needs memory.
Evidence needs witness.
Consequences need human gates.
Public page:
https://ivankotov.eu/publications/instrumental-c-public-portfolio/
GitHub:
https://github.com/Kot141078/instrumental-c-public-portfolio
Zenodo DOI: