The state of enterprise AI report
OpenAI released an info-rich report on the state of enterprise AI - here's everything you need to know
OpenAI’s new enterprise report is basically a progress update on a thing you can probably already feel at work: we’re leaving the “cute prompt experiments” era and entering the “AI is everywhere” era.
The headline is that usage intensity is exploding, and the teams getting real wins look a lot like the teams who got good at cloud. These teams standardized workflows, wired in internal context, and started measuring (the right) things.
Here are the highest-impact takeaways for software engineers and engineering leaders.
Enterprise AI usage is quickly becoming the norm
OpenAI reports that ChatGPT workplace adoption is now massive, with more than 7 million workplace seats, and ChatGPT Enterprise seats up ~9× year-over-year.
The most interesting part to me is how the usage is changing:
Weekly Enterprise message volume grew ~8× since November 2024, and the average worker is sending ~30% more messages.
Custom GPTs and Projects are becoming the “workflow packaging” layer with weekly users of these are up ~19× year-to-date.
The center of gravity in AI adoption is shifting from “prompting” to “building a library of small, dependable, repeatable assistants” that map to real workflows. I’ll probably be writing about this more this quarter
Power users are creating huge gaps
This report is unusually blunt about a reality most orgs try to hand-wave: even with broad access, a small set of power users do most of the serious work.
“Frontier” workers (95th percentile) send ~6× more messages than median users.
The gap is biggest for coding. Frontier users send ~17× as many coding-related messages as the median.
Even among active Enterprise users, there’s still notable feature underuse. 19% have never used data analysis, 14% never used reasoning, 12% never used search (monthly active).
Impact is scaling with depth of use
The report has a few big claims from aggregated usage data and a survey across 100 enterprises:
75% of surveyed workers say AI improved the speed or quality of their output.
Average reported savings of 40–60 minutes per active day (with some functions like data science, engineering, communications reporting 60–80 minutes).
Function-level “felt impact” is high. 73% of engineers report faster code delivery.
