For teams who were handed a login
Everyone has access.
Almost nobody uses it.
Because “here is a login” is not training, and nobody wants to look slow in front of colleagues.
How adoption actually fails
It fails in human ways, not technical ones
Licences bought for everybody
The easy part, and the part most companies stop at.
Most people try it once
Then go back to what they know works.
Very few are still using it a month later
“I didn’t know what to use it for.”
The real reason
Nobody ever showed them their own work being done differently.
The short answer
Adoption fails for human reasons, not technical ones. People will not experiment in front of colleagues, nobody wants to be visibly slow at something new, and a practice with no owner reverts within weeks. Buying licences addresses none of that.
Which is why the fix is a room where the first person to get a bad result is the person running the session, working on tasks the team actually resents doing.
- Slack surveyed 17,372 workers: 30% had received no AI training at all, and 61% had spent under five hours learning it.
- 45% did not have explicit permission to use AI at work — and 48% would be uncomfortable telling their manager they had.
- The reasons given were social, not procedural: that it feels like cheating, or looks lazy. Company policy was the least-cited reason, at 21%.
- Which is why licences change nothing on their own, and why any new practice without a named owner reverts once the training ends.
What we do instead
Their work. Live. In the room.
Each person names a task they resent
That is the first thing we try. Not a demo task.
We do it together, live
Including the attempts that do not work.
The result is judged honestly
Sometimes it is worse, and saying so builds more trust than a polished demo.
What worked becomes a shared way of working
Written down, not remembered.
Someone owns it after we leave
Any new practice with no owner reverts within weeks.
Nobody experiments publicly
The unspoken part
People will not experiment in front of colleagues
That is the actual barrier, and no amount of licensing solves it. It is solved by a room where the first person to get a bad result is the person running the session.
Evidence
What the research actually says
Every figure carries its source, date, the population it was measured across, and whether it is an observation or a projection.
30% of desk workers had received no AI training at all and 61% had spent under five hours learning to use it. 45% did not have explicit permission to use AI at work, and 48% would be uncomfortable telling their manager they had used it.
estimated Slack Workforce Lab, Workforce Index
12 November 2024 · 17,372 desk workers across 15 countries, surveyed independently of Slack customers · fieldwork 2–30 August 2024
The reasons people gave were social, not procedural: feeling that using AI is cheating (47%), fear of seeming less competent (46%) or lazy (46%). Company policy was the least-cited reason, at 21%.
Among firms that had adopted AI, 57% used it in three or fewer business functions, and 66% used it solely to augment existing tasks rather than replace them.
estimated US Census Bureau, CES Working Paper 26-25, “The Microstructure of AI Diffusion”
April 2026 · more than 117,000 distinct US firms · reference period November 2025 – January 2026
Three credible surveys of the same economy put AI adoption at roughly 18%, 41% and 78%. The Federal Reserve attributes the spread to sampling, unit of analysis, question framing and what counts as real usage — and notes that “adoption among the smallest firms is stronger than would be expected based on size alone”.
observed Federal Reserve, FEDS Notes, “Monitoring AI Adoption in the US Economy”
3 April 2026 · comparison of the Business Trends and Outlook Survey, the Real-Time Population Survey and the Atlanta Fed Survey of Business Uncertainty · published April 2026
Who does this
What SidRatnam.com does with your team
SidRatnam.com runs working sessions on the team’s own real tasks, live, including the attempts that fail. What works is written down as a shared way of working and handed to a named owner inside the business before we leave.
What that includes: each person naming a task they resent, attempting it together in the room, honest assessment of the result, and a short written reference the team keeps.
What it contributes to: a small number of tasks that genuinely change, and stay changed.
Live today
Connected problems
These turn out to be the same conversation
How can AI transform my business?
How do I stop doing admin at night?
How do I get my business to show up in ChatGPT?
This sits under transform my business with ai.
Make it stick this time
The free analysis looks at what your team actually spends time on before anyone is asked to change how they work.