Learning from use
Proposes a governed learning loop that turns user corrections and usage signals into improvements across knowledge, search, instructions, memory and model training, with evaluation and human approval before release.
Read the paperRESEARCH / ENTERPRISE AI
Governing enterprise AI agents
A framework for delegating platform administration to AI agents, covering identity, knowledge, connectors and audit trails, with human approval and strict access controls around consequential changes.
Read on this pageAt a glance
As an enterprise AI platform adds more users, knowledge sources and agents, administration becomes a workflow of its own. This paper examines how agents could help with identity, knowledge onboarding, agent curation, usage oversight and audit work while remaining subject to human authority.
Using Corporate GPT as its reference platform, the paper describes the governance interfaces and guardrails needed for delegated administration. It also considers agents that recognise recurring work and propose a purpose-built workflow for review. The focus is on how to bound the delegation: administrators need to see what would change, retain approval over consequential actions and be able to trace what happened. The paper separates that controlled direction from unrestricted autonomy over the platform.
What the paper covers
Platform administration is treated as a set of distinct, governed agent tasks.
Least privilege, reviewable changes and audit records constrain consequential actions.
Recurring work patterns can inform proposals for new agents or workflows.
Research context
The paper combines reference-platform capabilities with a framework for delegated administration and future autonomy. It does not establish that every proposed administrative agent is deployed or that the platform can administer itself without human review.
THE ORIGINAL WHITEPAPER
Download PDF Read it here.
The complete paper, with its original figures, tables and references.
19 pages · Read without leaving this pageFROM RESEARCH TO YOUR WORK
Discuss this research Bring your questions.
Discuss the approach, evidence and relevance to your team with the people building it.