⬡ Exhibit A

Uber blew through its entire 2026 AI coding budget by April.

Not from one bad decision — from an internal push to get engineers using AI as much as possible, tracked on leaderboards ranking staff by usage. Adoption soared. Then Uber's own COO admitted there was no clear line yet between all that usage and actually better products for riders and drivers. There's a name for it now: tokenmaxxing. Usage went up. Nobody had defined what "working" meant.

A five-person firm with eleven AI subscriptions and four active users made the exact same mistake Uber made. So did the marketing manager who signed up for three overlapping tools because each one looked good in a demo. Scale doesn't change the error. It just changes how many zeros are attached to the invoice.

The fix isn't "use AI less" or "spend less." It's sequencing: name the metric before you name the tool. If you can't say what success looks like in 30 days, you're not ready to buy anything — no matter how big or small the budget is.

This made the list because it's built around the metric, not the usage. AdCreative.ai generates ad creative — banners, social ads, product shots — against a stated goal: click-through rate, conversion rate, scroll-stopping performance. Every asset gets scored before you spend a dollar putting it in front of an audience.

That's the model this whole issue is about, in one tool: define what "working" means first, then let the spend follow the number — not the other way around.

📬 Tomorrow

Wednesday Use Case: a real team, a real stack, a real before-and-after.

Haven't named your 30-day metric yet? Talk to Jordan free → thepromptory.com