More token spend doesn't mean better engineering. In this session, we'll show you what productive AI work actually looks like and give you a framework to tell the difference starting immediately.‍

The problem

‍Your organization is sending three signals at once: use AI more, don't blow the budget, and show us what it actually produced. Most engineering teams have no way to answer the third question — and that gap is getting expensive.Token spend is easy to measure. AI value is not. This session is about closing that gap.

What we'll cover

Why "more AI usage" is not the same thing as better engineeringThe case for separating AI activity from AI value, and why your current metrics probably can't do it.

A live demo: same task, two approaches, real token costs

We run the same work two ways — one that burns tokens and creates cleanup, one that's structured and context-aware — and show you the cost difference. Then we do it again for a second team with a different type of work, so you can see how productive and wasteful AI patterns manifest differently depending on what your engineers are building.

A taxonomy for classifying AI spend

What productive, exploratory, and wasteful work actually look like in practice, and how to use that classification to make decisions about where to invest and where to cut.

Connecting token spend to real engineering work

PRs merged, tickets closed, features shipped, results delivered. What it takes to make AI spend meaningful at the leadership level, and what to do when you can't make that connection.

‍What you'll leave with

• A framework for classifying AI work into productive, exploratory, and wasteful spend.
• Practical patterns for structuring AI work to produce better outcomes that you can implement today.
• A model for tying token budgets to PRs, tickets, features, and results your leadership actually cares about.
• Language you can use with your team and your leadership to talk about AI ROI with precision.