OpenAI quietly dropped a number earlier this year that deserves more attention than it got. Over 5 million paid business users on ChatGPT, across organisations of all sizes. That's not free-tier hobbyists. That's companies paying $25 to $30 per user per month because the productivity gains justify the line item.
But the number I find more interesting: business users of AI tools report saving 40 to 60 minutes per day on average. That's roughly five hours a week. Per person. At a 20-person firm, that's 100 hours of recaptured time every single week.
Who's Actually Using This Stuff
There's a persistent myth that AI adoption is concentrated in tech companies and startups. It isn't. The fastest-growing segments are in professional services: law firms, accounting practices, consulting shops, financial advisors, and insurance brokerages. These are relationship-driven businesses where the work involves a lot of reading, writing, summarising, and communicating. Exactly the tasks where current AI tools are strongest.
A mid-size law firm in Toronto told me their associates now draft initial case summaries using Claude, then review and refine. What used to take 90 minutes takes about 25. They're not replacing anyone. They're getting more throughput from the same team.
The Social Proof Problem
I talk to a lot of SMB owners who are "sceptical but curious." They believe AI probably works, but they haven't seen it in a business that looks like theirs.
That's why the ChatGPT business number matters. Five million paid users means this has moved past early adoption. Gartner reported that 55% of organisations that piloted generative AI have moved to production deployments. The pilot phase is ending for companies that started. The ones that haven't are falling behind both early adopters and fast followers.
The Time Audit
Here's something you can do this week that costs nothing.
Pick four or five people on your team. Ask each to track their tasks for one full week. Just a rough log: what they worked on, how long it took, and whether the task was primarily reading, writing, researching, data entry, or communication.
At the end of the week, look for the patterns. Where is the most time going? Which tasks are repetitive? Which ones involve translating information from one format to another, like turning meeting notes into action items, or converting a client conversation into a formal email?
Those translation tasks are where AI delivers the fastest returns. Starting from a 70% draft and editing it down takes a fraction of the time that starting from zero does.
Most SMBs find at least one hour per person per day of AI-eligible work in their first time audit. That's your business case, right there.
What the Early Adopters Got Right
The businesses I've seen succeed started small. One workflow, one team, one tool. They gave it a real shot for a few weeks. And they measured honestly, including the failures.
They also didn't wait for perfect conditions. The tool wasn't perfect. The prompts weren't perfect. They started anyway. Six months in, their team has genuine competence that compounds daily.
McKinsey estimates that generative AI could add $2.6 to $4.4 trillion in annual value across industries globally. Zoom in: if AI saves each of your 15 employees 45 minutes a day, and your average fully-loaded labour cost is $40 per hour, that's roughly $180,000 per year. For a tool that costs $6,000 annually in subscriptions.
The math works. The only variable is whether you decide to start.