AI Governance is critical to the safety and security of your organisation
- mickbrawn
- May 25
- 2 min read
I spent four years at Microsoft managing their AI adoption consulting team across Asia - an almost absurdly massive geographical area, covered by a smallish, but consummately professional consulting delivery team. We discovered something important about the adoption of the Microsoft Copilot which I expect is applicable to the adoption of almost any LLM.
The adoption of many technologies (ERP/General Ledgers) requires highly capable technology teams, but the adoption of AI LLMs doesn't. Our adoption change management team was often more effective and drove faster and broader adoption that technical consulting teams. This is because there is very little technology effort in implementing Copilot – it is designed for secure integration into the organisation’s back-end systems and data - shifting the adoption challenge to solving the needs and desires of AI users.
The adoption of AI Agents is different. End users can certainly create their own AI agents for specific business processes but there is ample space for technologists in developing an organisational AI Agent infrastructure.
The challenge has now shifted to governance. Early adoption of any successful technology drives a period of rapid, unmonitored and uncontrolled expansion leading to massive duplication of effort and redundancy of compute power and of energy consumption.
A recent post on LinkedIn addressed this issue. Here’s a snippet:
“AI coding agents only ever get bigger. You add a skill. Then another. Six months later your agent is carrying 63 capabilities, using a fraction of them, and every unused one is dead weight...” The post explained that Copilot Agents Dojo v1.1 managed agents' capabilities & functions as active → stale → archived - driven entirely by real usage. Unused functions. Used functions revive instantly. Every agent backed up, reversible, and logged, and skills are protected from ever being touched.
This approach addresses governance early, before it becomes a real challenge.
The Boardpro organisation* (www.boardpro.io) recently “surveyed 485 board members, CEOs, and governance professionals across AU/NZ about AI adoption and AI governance.”
They found that
· “79% use AI weekly or more
· Fewer than 5% have a governance framework for it
· 75% of boards receive AI information ad hoc or never
· And 25% have not discussed AI-specific risks at all”
AI Governance is critical to the safety and value of your investment in AI.
The Australian Government established the National AI Centre to provide copious advice and guidance on implementing AI safely and securely, starting with governance. Take a look at their AI adoption insights: Dec 25 to Feb 2026 | National AI Centre.
I am happy to offer a free one-hour meeting with you to discuss AI governance, safety and security.
Mick Brawn
0414 987 129
Comments