A half-day session for learning and transformation leads
88% of employees use AI at work. 5% use it in ways that change the work.
Your rollout did not fail. It split. A small group worked out how to get real leverage from these tools, and everybody else settled for a faster way to look things up.
Half a day with the people who worked it out, and a written playbook of what they do for everybody who has not. No pre-work, nothing to prepare, and nothing confidential to send me.
Why the split happens
Three things, and none of them is effort
The people who plateaued are not lazier or less capable than the ones who did not. They were given a tool, a policy and a login, and left to work out the method on their own. Some did.
The tool became a search box
Most people stopped at the first answer
Ask a question, read the reply, move on. It works well enough to feel like using the tool, and it is the ceiling for the majority of an organisation. The people who broke through it did something different and unglamorous: they stopped asking questions and started handing over work, with the context and the constraints attached.
Nobody wrote down what worked
The knowledge is inside your building and it is trapped
Somewhere in your organisation are people getting real leverage out of these tools. What they know is specific to your work, your documents and your approvals, which is exactly why no course sells it. It lives in their heads, it spreads to whoever sits near them, and it stops there.
The training was aimed at the wrong rung
A curriculum teaches the tool, not the job
Vendor programmes are built to scale to a hundred thousand people, so they teach the product: the interface, the features, the safe use policy. That is worth something and it is not the gap. The gap is between a person who knows what the tool does and a person who has changed how they work.
The half-day
Three things, worked on real tasks
The room is your most effective users, the people who lead their teams, and whoever owns enablement. We work through tasks those people brought with them, which is what makes the output usable by everybody else.
One
Giving work away, not asking questions
The difference between a person who queries a model and a person who delegates to one. What can be handed over whole, what has to be broken up first, and what should never leave a human at all. This is the framing the rest of the day sits on, and it is the part that changes behaviour on Monday.
Two
Context is the job
Your best users are not writing cleverer sentences. They are supplying more of the situation: the document, the constraint, the example of what good looks like, the thing that must not happen. We take real tasks from your own people and rebuild them with the context that was missing.
Three
Judgement at speed
Output now arrives faster than anyone can read it, so the checking has to get cheaper and more specific. What is worth verifying, what can be spot-checked, what has to be read line by line, and how to recognise the confident, fluent, wrong answer that costs somebody a week.
Three things it deliberately does not cover: cognitive load, trust and empathy, and what a person is worth once everyone can generate a decent draft. Those are real questions and somebody better qualified than me should answer them.
What you keep
A playbook written from your own people
Fifteen people in a room does not change an organisation of two thousand. The document does. It arrives within a week of the session, it belongs to you, and the person who owns enablement can hand it out and run the material again without me.
- Eight to twelve worked patterns taken from your own people, on tasks your organisation actually does
- The context each pattern needs, written out, so somebody else can follow it without asking
- A verification checklist that matches the risk of the work rather than a generic one
- The three habits worth dropping, named plainly, because stopping is faster than learning
- Everything in your language and your file formats, so it can go straight into your own learning platform
How it runs
Three conversations, half a day, then the document
Once we have agreed the work, I speak to three or four of the people you have identified as getting the most out of these tools. Half an hour each, about how they actually work. That is where the patterns come from, and it is why the session is about your organisation rather than a prepared example.
Then the half-day itself, remote or on-site, up to fifteen people. Then the playbook, within a week.
Nothing confidential leaves your organisation. The conversations are conversations, no exports, no documents sent to me, no access to your systems. Attendees prepare nothing and bring nothing except the work they were going to do that week anyway.
Where the numbers come from
One survey, and one number I threw out
88%
of employees use AI at work, in EY's words "primarily limited to basic tasks such as search and summarization".
EY Work Reimagined Survey, August 2025
5%
are "using it in advanced ways to transform the way they work". The same tools, the same access, the same organisation.
Same survey: 15,000 employees, 1,500 employers, 29 countries
40%
of the available productivity gain is being missed, which EY attributes to gaps in talent strategy rather than to the technology.
EY newsroom, November 2025
A figure doing the rounds is that power users are four times more productive than everybody else, credited to OpenAI. Their report does not say that. It reports frontier workers sending roughly six times more messages, which is a measure of activity rather than of output, and the two are not the same claim. It is not on this page.
Right for
- Organisations where the tools went out at least three months ago and usage has settled
- Anyone who can name the handful of people getting real value, and cannot say what they do differently
- Learning and transformation leads being asked what the licences bought
Wrong for
- Organisations that have not rolled anything out yet, where the useful conversation is a different one
- Engineering teams building agents, who are better served by the Agent Workshop
- Anyone wanting a tool recommendation. The session is provider-agnostic and works with whatever you have already bought
Who runs it
Someone who works this way daily
I am Tamas Piros. Twenty-five years across software engineering, developer relations and technical training, a Google Developer Expert in Web Technologies and an ambassador for the Agentic AI Foundation. 138 talks and workshops in 33 countries, and a free book on building with these tools that you can read right now without giving anyone your email address.
The reason that matters here: extracting a working method from somebody and writing it down so a stranger can follow it is the job I have done for twenty-five years. The people in your building who cracked this are usually poor at explaining it, which is exactly why it has not spread on its own.Judge the teaching yourself, before we ever speak.
Start
Tell me when the tools went out and who is getting the most from them. I will tell you whether this is the right thing and what it costs.
Teams who want the practice to stick run this as three sessions across six weeks, with real work in between. Worth talking about after the first one rather than before it.
Some organisations keep me on afterwards for the questions that arrive once people start working differently.