Do you trust your coding agents (and the budget they spend)? - FairMind Breakfast
AI coding agents now write more code than any team can read carefully. Somewhere between the pull request and the merge, someone has to decide whether that code can be trusted, and today the answer is almost always to re-read all of it by hand. That is where the budget quietly goes, and it is what we want to talk about over breakfast.
The issue is rarely which model you choose, it sits instead in what happens before an agent reaches its objective: loops that restart from zero, tasks that drift and need to be redone, decisions the agent has no memory of because the session that made them already closed. Each of these costs twice, once in tokens and once in review hours, and better models will not close the gap on their own, because verifying that code still respects a decision made months ago is not something a model gets rewarded for during training.
Three things we want to get into, over coffee:
Trust & Verification – how teams decide what an agent's output still needs a human eye, and what does not
Loop Engineering & Best Practices – stop conditions and verification gates that catch drift before it compounds
Cost Governance & Tracking – token consumption and review hours, tracked by team and project
This breakfast is for CIOs, CTOs, CEOs, COOs, Heads of Innovation, Heads of Transformation, and anyone leading operations within a software factory. Or simply for developers!
Join us for breakfast.
PS not all the info in the registration form are mandatory, but sharing who you are will definitely improve your experience allowing us to personalize the event. Also, being a closed event enhances the chance of being accepted :)