OpenAIGuides

Inside GPT-5 for Work: How Businesses Use GPT-5

January 22, 2026 at 12:00 AM

Summary

TL;DR

ChatGPT’s workplace adoption is being pulled in bottom-up from consumer use, then formalized by enterprises. Early usage concentrates on low-friction tasks, while advanced capabilities are mostly used by technical roles.

What actually happened

OpenAI combined third-party studies with anonymized/aggregated ChatGPT and ChatGPT Enterprise usage analysis.
Workplace adoption rose quickly after consumer uptake; employees “ported” usage into jobs without long onboarding.
In enterprise first-90-days data, most usage falls into writing, research, programming, and analysis.
Technical departments use more advanced tools; most other roles stick to core features.
Some industries (IT, finance) adopt faster; others (retail, construction, transportation, agriculture) lag.

Key numbers

28% of employed adults use ChatGPT at work (up from 8% two years ago).

Over a quarter of U.S. workers report using ChatGPT for work.

45% of workers with postgraduate degrees report using ChatGPT for work.

43% of U.S. knowledge workers use AI (up from <10% in late 2022).

ChatGPT: 100M weekly active users within months; now 700M+ weekly active users.

Some ChatGPT Pro power users send 200+ messages per day.

AI users: over half save 3+ hours per week (Federal Reserve Bank of St. Louis study).

Knowledge workers using AI produced 40% higher quality work (Harvard study).

Randomized field experiment: AI cut weekly email time by 31% over six months.

Employees 18–29 are more than twice as likely to use ChatGPT at work as those 50+.

Why this was hard

Adoption varies widely by industry; sectors with fewer knowledge workers move slower.
Healthcare adoption is constrained by strict privacy/compliance and risk-averse culture.
Advanced features are underused, potentially due to discoverability and setup friction.
Different departments need different tool mixes (e.g., IT troubleshooting vs engineering coding workflows).

How they solved it

Analyzed usage on anonymized/aggregated data; did not review user/customer content.
Used automated content classifiers to categorize usage trends and first-90-days patterns.
Mapped professional email domains to industries to compare adoption by sector.
Tracked tool usage by job category (e.g., search, data analysis, file uploads, retrieval).
Positioned GPT-5’s real-time router to auto-select tools/features by conversation complexity and intent.

What changed

ChatGPT shifted enterprise adoption from top-down rollouts to bottom-up experimentation and proof-of-value.
Use expanded from Q&A into coding, data analysis, and agentic workflows.
Coding usage spread beyond engineering into roles like design and project management.

Why this matters beyond this company

Consumer-led adoption can bypass classic enterprise onboarding, accelerating workplace tool standardization.
If advanced features are hard to find or configure, most users stay on “core” workflows.
Department-specific needs drive tool choice: retrieval/search for IT, coding/data analysis/GPTs for engineering.

Stealable ideas

Measure first-90-days adoption by capability buckets (writing, research, programming, analysis).
Use anonymized, classifier-based telemetry to spot underused features and training gaps.
Segment enablement by department: IT troubleshooting vs engineering build/debug vs go-to-market content.
Track “messages per user” to identify power-user cohorts and their workflows.

Read original article

Inside GPT-5 for Work: How Businesses Use GPT-5

Summary

What actually happened

Key numbers

Why this was hard

How they solved it

What changed

Why this matters beyond this company

Stealable ideas

More from OpenAI

Inside GPT-5 for Work: How Businesses Use GPT-5

Summary

What actually happened

Key numbers

Why this was hard

How they solved it

What changed

Why this matters beyond this company

Stealable ideas

More from OpenAI