The Real Cost of AI Coding - Finding and Fixing Hidden Token Waste

AI coding tools are quickly becoming core infrastructure for engineering teams, but the way most companies manage their spend hasn’t caught up. As Claude Code and Codex usage scales across an organization, costs can grow for reasons that have little to do with how much useful work developers are getting done. Bloated context, unused MCPs and skills, misconfigured compaction, and inconsistent setups can quietly consume a significant share of the budget. Join Jacques Verre, Head of Product at Comet, to dig into what engineering teams should actually understand about their AI coding spend. He’ll cover how to connect costs back to engineers, teams, and the work being done, where hidden token waste tends to come from, and how teams can reduce unnecessary spend without restricting how developers use AI.