Five principles for AI training that helps employees do more of their best work—and helps organizations evolve with them.
I chose English Literature for my A-Levels because I thought it would be easy. Identify the themes of a novel, explain them in an essay, collect good marks. Throw in something about the weather reflecting a character’s mood and you were practically guaranteed an A.
My first few grades were D, C, D.
“So what?” “Why is this interesting?” “This reads like a critic, not like Cameron.” These were some of my teacher’s margin scribbles.
I asked what she meant. Wasn’t understanding the novel, then explaining it to the examiner, the whole point?
“Unlike lesser subjects,” she said—yes, she was a little like Robin Williams’ character in Dead Poets Society—“meaning is not learned, it is created.”
Reading, she argued, was participatory. I needed to bring my experiences, opinions, and assumptions to the text, and see what meaning emerged.
That changed my enjoyment of the subject and my understanding of learning. My own perspective had a legitimate place in the learning process. What I brought to a book’s reading shaped what I took from it, and how I applied what I learned from it back into my own life. I went on to study English at university and still read fiction voraciously.
We’re all in learning mode again. AI is asking us to learn new tools and reconsider how we work. Following instructions can teach us a useful technique. But if we want AI learning to change what people are capable of, they need room to bring their own experience and expertise to the learning process.
That’s the method I bring to Braid’s learning programs: helping people take an active role in discovering what AI could mean for their work.
Five principles that shape how I approach AI learning:
1. Design learning around people’s best work
One of my favorite questions before starting a learning program is: which part of your work would you like to do more of and what prevents you from doing more of it?
You hear about what gives people energy and where they feel they make their strongest contribution. That’s a useful starting point for learning, especially when those ambitions connect to something the organization needs more of.
Someone might be brilliant at understanding customer needs but spend most of their week assembling reports. AI could help them explore customer feedback more deeply or challenge their assumptions before a conversation. It could also help prepare those reports more quickly, creating more time for customers. Both serve the same ambition: helping that person do more of their best work.
This takes more preparation than delivering a standardized curriculum. Some ambitions can be addressed through shared sessions; others warrant individual attention. The learning practitioner’s job is to connect what people want to achieve with where their learning could create meaningful value for the organization.
People have a reason to put effort into a learning program when they can see what it makes possible for them. You’re pushing on a door that’s already opening, while helping the organization benefit from more of what its people have to contribute.
Put it into practice: Before designing a learning program, ask people what they want to do more of, what gets in the way, and how doing more of it would benefit the organization.
2. Teach people to redesign their own work
We tend to experience work as a whole: prepare the proposal, plan the event, write the report. We’ve done these things so often that the individual steps become invisible.
But a workflow like preparing a proposal involves multiple tasks: understanding a brief, gathering evidence, developing a recommendation, testing the argument, presenting it. Each task requires different information, skills, and decisions and so AI’s contribution during each task will vary accordingly.
At Braid, we teach a work-design practice called Untangle & Braid. You can read more about the method here.
First, we untangle a workflow into its tasks and ask what good work looks like at each stage. Then we decide how human and AI contributions could braid together. AI might collate information; a person might check that information with colleagues. Some tasks may have no useful role for AI. Some might be fully AI.
Looking at the work this way can also reveal steps that could be combined, handoffs that are no longer necessary, or opportunities for someone to own more of the overall process.
Then we test the approach on a real workflow, judge the result, and refine it. That repetition is how people learn how to design their work.
This gives people a practical way to investigate how to meaningfully integrate AI into their own work. By teaching people how to perform simple work design, organizations are able to scale workflow improvements: people are empowered to enact change in their own workflows rather than wait for an organization-wide change initiative.
Put it into practice: Apply the Untangle & Braid approach by taking one recurring workflow, breaking it into tasks, and identifying the possible role for AI and the human at each stage. I’m happy to share workshop templates to use, just email me.
3. Help people discover they’re newly capable of with AI
“Practical use cases” can become an unnecessarily narrow brief for effective learning. Relevance matters, but people can draw useful lessons from exercises that stretch them beyond their everyday responsibilities.
Scott Kirby, CEO of United Airlines, described building an experimental betting website with AI as “eye opening to me.” Building a betting website isn’t a practical use case for a CEO. But it’s an exercise that provides inspiration and confidence to explore what’s possible with these new tools.
I’ve worked with clients to create miniature video games during learning programs. The realization that they’ve made a functioning game can be as valuable as the skills they used to build it. Its greatest value was changing people’s expectations of themselves.
The most effective stretch task will differ from person to person. And once it’s over, we need to help them connect that experience back into their work. What did they learn? Where else could they apply the same approach? What might they now feel ready to try?
Learning programs must leave room for this kind of curiosity-led learning.
Put it into practice: Alongside everyday applications, include a supported experiment that stretches people beyond their current confidence. Finish by asking what they learned and where they could apply it in their own work.
4. Connect different ways of learning
A workshop, an online course, a one-to-one coaching session each offer a different learning experience. Group sessions can establish broad foundations. Independent learning lets people explore at their own pace. One-to-ones make room for questions someone might hesitate to ask publicly.
These experiences become more valuable when they build on one another. Someone might encounter an approach in a group session, try it in their own work, and bring a disappointing result to a one-to-one coaching session. The group session gives them a starting point, practice reveals where they struggle, and coaching helps them understand what to change. They return to the same problem with a deeper understanding each time.
That’s where embedded learning expertise can be incredibly valuable. A practitioner who knows the people and the business can connect these experiences, recognizing when someone needs another explanation, focused practice or a more ambitious challenge. Each interaction builds on what came before and helps shape what comes next.
In Braid’s programs, we combine group sessions, office hours, one-to-ones and peer exchange. Sometimes we lead; sometimes we help colleagues learn from one another. The aim is to build a culture where people have the time, support and confidence to keep learning together.
Put it into practice: Build a regular rhythm of shared sessions, independent practice, individual support and peer exchange. Give someone in your organization responsibility and time to “own” the learning agenda: it’s an involved, high touch role. You can call us if you don’t have that capability in-house.
5. Evolve the organization as people become more capable
Someone may learn to take a piece of work much further, while the process still requires them to hand it off at the same point. A team may find a better way to reach a decision, while approval remains several layers away. Learning has created new capability, but the organization’s current design leaves that capability unused.
As people’s capabilities change, leaders need to reconsider which organizational design decisions still serve the work and which now get in the way. Otherwise, they risk frustrating newly energized people and leaving the value of their learning unrealized.
Learning should therefore inform decisions about workflows, roles, knowledge access and responsibility. The people applying new capabilities are well placed to explain what needs to change around them: leaders need to listen.
This is why I see learning as an engine of organizational development. People need enough practical experience with AI to help shape a different way of working, and the organization needs to evolve those ways of working so their people can put that experience to use. Learning and organizational development go hand-in-hand.
Put it into practice: Throughout the learning program, ask people “what needs to change here for you to apply what you’ve learned?” Remove those existing barriers as fast as possible so that new learning can be applied in short order.
I want people learning AI to feel agency: to bring what they know, discover what else they can do, and have a say in how their work changes as a result. That’s how learning becomes a force for organizational change.
