Efficiencymaxxing
That’s the productivity paradox in one stat. The tool works, the leadership doesn’t.
Why “Using AI More” Was Never the Goal
Hey, it’s Samet 🙌
If your Slack has turned into a leaderboard of who’s burning the most AI tokens this week, I need to tell you something: that phase is already dying.
The smartest product teams are moving on.
What Efficiencymaxxing Actually Means
“Maxxing” is the internet’s favorite suffix right now. Attach it to anything and it means optimizing that thing to an obsessive degree, from sleepmaxxing to looksmaxxing.
Efficiencymaxxing is the AI-era version. Instead of chasing raw AI usage volume, you’re optimizing for output quality per unit of cost and time.
Every‘s Laura Entis put it perfectly this month: for a brief window, AI was cheap and power users proved their status by simply maxing out token consumption. That window has closed.
The Tokenmaxxing Hangover
Here’s the part every PM needs to internalize: companies actively incentivized the wrong behavior, and it backfired hard.
The Financial Times reported Amazon employees were “tokenmaxxing,” gaming internal AI usage metrics, while Meta ran an AI leaderboard to push adoption.
Uber reportedly burned through its entire 2026 AI coding budget in four months. Microsoft has been pulling back Claude licenses because, in Nvidia VP Bryan Catanzaro’s words, “the cost of compute is far beyond the costs of the employees”.
BCG’s 2026 Global AI at Work report, surveying nearly 12,000 frontline workers, found the real cost of this misfire: 42% of employees saved a full workday a week using AI, but 66% got zero guidance on what to do with that saved time, and half never redirected it toward anything strategic.
That’s the productivity paradox in one stat. The tool works, the leadership doesn’t.
From Volume to Intent
The shift happening right now, and the one you should be building your roadmap around, is a move from “how much AI are you using” to “what can you actually show for it”.
A few signals worth tracking:
Revenue per million tokens is emerging as a new efficiency metric, spotted circulating in San Francisco tech circles at Apple’s WWDC. It’s a successor to revenue-per-employee that ties AI spend directly to ROI
Amazon has already scrapped internal AI-usage tracking after staff deployed bots to game the metric, with a senior VP telling staff plainly: “Please don’t use AI just for the sake of using AI”
BCG’s People & Organization lead David Martin argues the fix isn’t more mandates. It’s leaders articulating a clear vision for why AI is being used, paired with real upskilling so employees stop hoarding AI tricks out of fear
What This Means for Your Product
If you’re building or managing AI features, efficiencymaxxing is your new north star metric, not adoption rate.
Ask: does this AI-generated output actually save downstream editing time, or does it just create more content to clean up?
Author Craig Mod’s approach is a useful gut check. He still vibe-codes SaaS alternatives with AI for research, but writes every word himself because outsourcing the “mess” would defeat the point.
Apply that logic to your team: automate the grunt work relentlessly, but protect the parts of the job where human judgment is the actual product.
The takeaway for this week: audit one AI workflow on your team and ask whether it’s optimizing for tokens burned or value delivered. If you can’t answer confidently, that’s your next sprint.
Talk soon,
Samet Özkale, AI for Product Power