The Sovereign Engineer: Making Agents Responsible for Human Understanding and Judgement
Part of the AI Literacy Series
(Available as a keynote and session talk, in-person and virtual,
conference or internal to an organisation)
Intelligent developer platforms promise to remove friction, cognitive load, and eventually the need to understand the systems we ship. That last one is not a feature.
Drawing on Peter Naur's insight that a program is the theory its builders hold of it, this session argues that as platforms absorb more of the work, the scarce resource is no longer delivery speed — it's human judgment, and the understanding that discerning judgment depends on. AI is an amplifier: it magnifies whatever engineering practice it lands on, including the quiet habit of surrendering comprehension one convenient step at a time.
Russ tells that story from the inside — the moment he realised he no longer understood what he was building well enough to trust his own decisions — and then shows the way back through "Level 5: Supervising" on the agentic software engineering ladder: making agents responsible for maintaining human understanding, not just for producing output.
Using open-source "sentinel" agents from the AI Literacy Superpowers plugin — an Advocatus Diaboli that adversarially reviews specs, a Choice Cartographer that surfaces decisions before they're silently made for you, a Carpaccio agent that enforces thin slicing, a Reservoir Warden that watches cognitive exhaustion, amongst others — the session demonstrates patterns any team can adopt today.
You'll leave with a working sentinel architecture and a sharper answer to the question every intelligent platform must eventually face: who is the accountable party, and how do we support them?