The March 2026 Systems Thinking Systems Practice conference in Hull included a presentation "Agentic AI and the Self-Managing Enterprise" [the slide set is included here] which explained that we - and AI agents - control a system by decisions that try to close a gap between some current state of the system and a target for that state. It went on to explain that, for AI agents to do that successfully across the enterprise, each needs to take account of the impacts of other agents' decisions. This presentation goes on to explain how that collaboration among AI agents needs to work. First, any local policy (e.g. for the number of staff we need to hire) must be moderated by 'guardrails' that prevent other unacceptable outcomes (e.g. too high staff cost). Secondly, the system needs a supervisory control level that mediates between policies that interact (e.g. do we hire more staff, or slow the growth that those staff are supporting?) The presentation will demonstrate how this is implemented with a live case.
Events associated with this resource
There are no events associated with this resource.