|
Monday · August 3, 2026 · Issue #042
Uber's CTO said it plainly this spring. "I'm back to the drawing board, because the budget I thought I would need is blown away already."
The company had rolled out Claude Code to roughly 5,000 engineers in December 2025. By April 2026 — four months later — the entire annual AI budget was gone. Not because anything went wrong. Because everything went right. Engineers loved it. Adoption jumped from 32% to 84% of the engineering org in months. About 70% of committed code now comes from AI. And nobody had modeled what that would cost, because nobody had ever modeled it before.
A separate enterprise, unnamed in the reporting, spent $500 million on AI services in a single month. The reason: nobody had switched on a usage cap for its Claude licenses. Unlimited access looked like flexibility right up until the invoice arrived.
This is the AI story nobody was talking about six months ago. The conversation has shifted from "how do I get my team to use AI" to "how do I make sure I'm not Uber." Today's issue is about that shift — and what it means specifically for professional services firms operating without a $3.4 billion R&D budget to absorb the lesson.
|
|
⯯ The Numbers Behind This Week's Story · All Named Sources
78% of IT leaders have encountered unexpected charges from consumption-based AI pricing. — Zylo 2026 SaaS Management Index
79% of enterprises experienced AI cost overruns in the past year. — 2026 survey of 500 finance leaders, US and UK
40% of AI agent projects will be cancelled by 2027 due to cost overruns alone — not technical failure. — Gartner 2026 AI Hype Cycle
Agentic AI uses 5–30x more tokens per task than a standard chatbot. — Gartner March 2026; Stanford Digital Economy Lab April 2026
One two-hour Uber session cost $1,200. Microsoft cancelled coding-agent licenses six months into its pilot. — PointFive research, Bloomberg
|
|
|
⯯ Why This Happened · And Why "We're Not Uber" Is Not a Defense
|
|
|
The Uber story is easy to dismiss if you're not a 5,000-engineer org. Don't dismiss it.
The mechanism that blew Uber's budget is the same one operating at every scale. AI tools have moved from flat-fee subscriptions to consumption pricing — every prompt, every agent action, every automated task now generates a cost that compounds. Sam Altman said it plainly in March: "We see a future where intelligence is a utility, like electricity or water, and people buy it from us on a meter." That future arrived faster than most finance teams were ready for. Gartner confirmed the structural shift: agentic models require 5 to 30 times more tokens per task than a standard chatbot — meaning the cost model that worked for ChatGPT-as-a-writing-tool breaks immediately when you deploy it as an agent.
The problem wasn't the tools. It was the absence of four things that should have existed before any tool was deployed at scale. A defined scope. A cost model built before deployment. Usage caps or spend alerts that fire before the invoice. And someone accountable for outcome, not just adoption rate.
Uber incentivized teams with a leaderboard ranked by total AI tool usage. They got exactly what they measured. Maximum usage. No accountability for what that usage produced.
|
|
⯯ The Incentive Problem · Worth Naming Directly
When you reward the metric of "using AI," you get maximum AI usage. That is not a technology problem. It is a management problem that technology made expensive.
Every professional services firm pushing AI adoption through encouragement and internal champions — without a corresponding system for measuring whether usage produces outcomes — is running a slower version of the same experiment. The cost is smaller. The structure is identical.
Uber's COO put it plainly in May: "It's very hard to draw a line between one of those stats and, 'Okay, now we're actually producing 25 percent more useful consumer features.'" That's the question that matters. Not whether employees are using AI. Whether that use is producing something measurable.
The organizations that avoided the budget crisis were not the ones that used AI less. They deployed it against a defined problem, with a defined cost model, measured against a defined outcome. Discipline beat enthusiasm. Every time.
|
|
|
⯯ Three Questions Your Firm Should Be Able to Answer Right Now
|
|
|
QUESTION 1 · What is your firm's total monthly AI spend — across every tool, every account tier, every team member?
Not the subscriptions you formally approved. All of it — including personal accounts used for work, free tiers that become paid when usage crosses a threshold, and any API costs from tools your team has connected together. If you cannot answer this in under five minutes, you have a visibility problem. Zylo research found an average of 18% of enterprise AI spend is unattributed — one dollar in five disconnected from any team, tool, or outcome. That ratio does not improve with smaller firm size. It just costs less per month.
|
|
|
QUESTION 2 · What specific outcome is each AI tool connected to — and what number tells you it's working?
Not "it helps with productivity." A number. Client onboarding time. Follow-up response rate. Hours on document review. Revenue per person. Something that existed before the tool and should be different because of it — and something someone is actually tracking. If the answer is "we haven't defined that," you're running Uber's experiment at your firm's scale. The cost is different. The accountability gap is the same.
|
|
|
QUESTION 3 · Who is accountable for AI spend — and do they see it before the invoice, or after?
Uber's answer to this was "nobody, apparently." Anthropic launched Claude Enterprise spend controls on July 2, 2026 specifically because enterprises kept hitting this wall — model-level entitlements, spend-threshold alerts, real-time analytics. The governance layer arrived because the market demanded it after enough budgets blew. For a professional services firm, this does not require enterprise infrastructure. It requires one person whose job includes knowing the answer, and one system that surfaces it before surprise becomes crisis.
|
|
|
⯯ Jordan · AI Solutions Director · thepromptory.com
Powered by Claude Opus 5 · Free · No account required · No sales call after
|
|
P
|
I read the Uber story and got nervous. We have 12 people and I genuinely don't know what our total AI spend is right now. I know we have Copilot, someone pays for ChatGPT Plus, and I think a couple of team members use tools I haven't formally approved. How do I get a handle on this without making it feel like a crackdown?
|
|
J
|
Good instinct on the framing — you want visibility, not surveillance. Start with a team conversation framed around building a better system, not catching anyone out. Ask everyone what AI tools they're using and roughly how often. Tell them you're consolidating to the best options and want to know what's actually useful. That gets you the full inventory without the crackdown energy. Once you have it, I can help you figure out which tools to keep, which to replace with better-governed alternatives, and what a sensible spend framework looks like for a 12-person firm. What industry are you in?
Jordan · thepromptory.com →
|
|
|
Not sure what your AI stack is actually costing? Jordan helps you find out → thepromptory.com
|
|
|
|
⯯ How The Promptory Is Built for This Moment
|
|
|
The Uber story marks the transition from the adoption conversation to the accountability conversation. Most AI tools, directories, and newsletters were built for the first conversation. The Promptory was built for both simultaneously — which is why the approach looks different.
|
|
Problem definition before tool selection
Jordan asks what specific problem you're solving before recommending anything. Uber's problem wasn't that Claude Code is bad. It's that nobody scoped what it was for, so adoption was unlimited by design. When the problem is defined first, the cost is bounded by the scope.
|
|
Curated tools at the right price tier for your actual use case
The vault vets more than 130 tools against a five-point standard — transparent pricing is non-negotiable. Every recommendation includes the actual cost at your usage level. The right tool at the wrong price tier is still the wrong tool. Curation prevents that mistake before it compounds.
|
|
Governance infrastructure built in, not added after
The Core System Build includes governance from day one — Airia for data controls and audit trails, the AI policy framework for what your team can and can't do, a defined outcome metric specified before anything goes live. These aren't add-ons. They're the difference between a system and a subscription.
|
|
|
|
💡 The One Thing
Uber's budget crisis wasn't caused by bad tools or reckless engineers. It was caused by measuring adoption instead of outcomes — and having no system to see the cost until it was already spent. Most professional services firms are running a version of this. The scale is different. The structure is identical.
Gartner forecasts 40% of AI agent projects will be cancelled by 2027 due to cost overruns — not technical failure, not market fit. Just the absence of discipline. The firms that avoid that list aren't using less AI. They're using it against a defined scope, with a defined cost model, measured against a defined outcome.
The meter is running. The question is whether you know what it's measuring.
|
|
|
📬 This Week
Tuesday through Friday: practical moves for the transition from AI adoption to AI accountability. The audit that tells you what your stack is actually costing. The metric framework that connects spend to outcome. And the specific vault tools that deliver governance without enterprise infrastructure budgets.
The meter is running. Know what it's measuring: thepromptory.com →
|
|