--- title: "Your AI rollout needs behaviour design | &above" canonical_url: "https://www.andabove.com/feed/your-ai-rollout-needs-better-behaviour-design" last_updated: "2026-10-06T14:54:10.897Z" meta: description: "You've bought the licences and run the training, but people still aren't using AI. Adoption is a behaviour change challenge, and it has to be designed for." "og:description": "You've bought the licences and run the training, but people still aren't using AI. Adoption is a behaviour change challenge, and it has to be designed for." "og:title": "Your AI rollout needs behaviour design" "twitter:description": "You've bought the licences and run the training, but people still aren't using AI. Adoption is a behaviour change challenge, and it has to be designed for." "twitter:title": "Your AI rollout needs behaviour design" --- # &above [&above](https://www.andabove.com/index.md) [Work](https://www.andabove.com/work.md) [Services](https://www.andabove.com/services.md) [Agent Mesh](https://www.andabove.com/agent-mesh.md) [Feed](https://www.andabove.com/feed.md) [About](https://www.andabove.com/about.md) [Contact](https://www.andabove.com/contact.md) --- # Your AI rollout needs better behaviour design `AI` *6 Oct 2026* ![Your AI rollout needs better behaviour design](https://www.andabove.com/content-media/feed/your-ai-rollout-needs-better-behaviour-design/hero.gen.jpg) You've bought the licences, you've run the training, you've told everyone AI is important. So why aren't people using it? The instinct is usually to look at the technology. Is this the right model? Is the tool good enough? Do people need more training? But there's another possibility: whilst you've been busy designing the rollout, you haven’t considered the behaviour. But AI adoption is not a one off implementation problem, it's a behaviour change challenge. ## Don't assume people will change because the opportunity is obvious One of the biggest mistakes businesses make with AI is assuming that because the technology is powerful, people will naturally want to use it - they won't. People are busy, they have established ways of working, they have deadlines, habits, preferences and concerns about getting things wrong. AI introduces another layer of uncertainty. What should I use it for? When should I use it? Can I trust the answer? Is my manager expecting me to use it? Is someone monitoring what I'm doing? Giving someone access to a tool doesn't answer any of those questions. A training session doesn't either. If you want behaviour to change, you need to design for it. ## Start with the behaviour, not the tool A useful place to start is with a much more specific question: What do we actually want people to do differently? "Use AI more" vs "Use AI to prepare for every sales call”. Once the behaviour is specific, you can start designing the conditions around it. That's where behavioural science becomes useful. ## 5 things that make AI adoption more likely ### 1. Give people a reason to care If AI is introduced as a way to monitor productivity, reduce headcount or simply make people work faster, it's hardly surprising if people are cautious. A better starting point is: how does this make someone's everyday work better? That might mean removing tedious administration, reducing repetitive work or giving someone more time for the part of their job that requires judgement. And show AI's weaknesses too. People need to understand where human judgement still matters. The goal isn't human versus AI, it's understanding where each is useful. That framing creates a very different relationship with the technology. ### 2. Design a cue, not another training session Habits need triggers. Think about something as simple as brushing your teeth. You don't attend a quarterly workshop on oral hygiene before deciding whether to do it. The behaviour is connected to a cue, a routine and a reward. AI rollouts often have the opposite structure: Training → licence → good intentions → nothing happens. There is no cue in the workflow telling someone when AI could help, there’s no protected time to experiment or any obvious reward when they do. If you want a new behaviour to stick, put the cue where the work happens. That could be a prompt in an existing workflow, a recurring team session, a manager demonstrating how they use AI, or a small task that people complete at the same time each week. The closer the cue is to the behaviour, the easier adoption becomes. Think Couch to 5K, not AI bootcamp. One of the best examples of behaviour design is Couch to 5K. It doesn't tell someone to suddenly become a runner. It starts small - a short run, a clear schedule, repetition, visible progress, a sense of achievement. And ideally, someone else doing it with you. The same principle can apply to AI. Instead of giving someone a two hour training session and expecting them to completely rethink how they work, start with one small, useful behaviour. Give them a task or a prompt and let them see the result. Then repeat it and increase the difficulty. And build from there. AI capability is built through use, not just instruction. ### 3. Make experimentation socially safe People look to other people when they're uncertain, that makes AI adoption surprisingly social. If nobody on your team appears to be using AI, using it can feel like a risk. If your manager openly uses it, shares what worked and talks about what didn't, the behaviour becomes more normal. This is why leaders matter so much. Leaders don't just communicate the AI strategy, they create the social permission to experiment. A manager who says, "We should all use AI" is sending one signal. A manager who says, "I used AI to prepare for this meeting, here's what it gave me and here's where I disagreed with it" is demonstrating another. The second is far more useful. ### 4. Give people time to play There's a contradiction at the heart of many AI rollouts. Businesses want people to experiment with AI but they expect them to do it on top of their existing workload. Then when usage is low, they conclude that people aren't interested. Sometimes they aren't but often they're just busy. Experimentation needs space to give it somewhere useful to go. ### 5. Build ownership, not just automation There's another reason to be careful about making AI too automatic. People value things more when they've had a hand in creating them. It's the principle behind the IKEA effect: we tend to place greater value on things we've put effort into building ourselves. The same idea can apply to AI. If an agent simply produces an answer, the user can become a passive recipient. But if the agent asks for context, challenges the user's thinking and shows how the result is improving, the person becomes part of the process. That matters particularly when the work requires judgement. The goal shouldn't always be: How much human work can we remove? Sometimes the better question is: How can AI make human judgement more valuable? That might mean designing an agent as a coach rather than an autopilot. It might mean adding check-in points before an important decision. Or it might mean asking the user to provide context before the system acts. Automation is useful where we don't need to think, where judgement matters, we should design for collaboration. ## …And then measure what happens Don't assume that the person or team using AI the most is creating the most value. Instead, we should look at the progression: Usage → Shared → Trusted Are people using AI regularly? Are useful workflows, skills and agents being shared rather than trapped in individual chat windows? Are people trusting AI enough to use it for deeper work? And is the quality of that work improving? That gives you a much better picture of whether behaviour is actually changing. Because adoption isn't the end goal, useful behaviour is. ## The AI rollout doesn't end when the training does The businesses that get the most from AI understand how people change behaviour, they make the desired behaviour easy to start, give people cues to use it, create space to experiment and make successful behaviour visible. Leaders model it and teams share it. And people remain involved where human judgement matters. That's designing an environment where using AI becomes a natural part of how work gets done, because you've designed the conditions that make people want to use it. ## Next up - [Your AI is only as good as the knowledge it can retrieve](https://www.andabove.com/feed/your-ai-is-only-as-good-as-the-knowledge-it-can-retrieve.md) - [Your AI pilot worked. Now who owns it?](https://www.andabove.com/feed/your-ai-pilot-worked-now-who-owns-it.md) - [View our feed](https://www.andabove.com/feed.md) --- ## Footer ### Pages - [Home](https://www.andabove.com/index.md) - [Work](https://www.andabove.com/work.md) - [Services](https://www.andabove.com/services.md) ### Studio - [About](https://www.andabove.com/about.md) - [Careers](https://www.andabove.com/careers.md) - [Feed](https://www.andabove.com/feed.md) - [Press kit](https://www.andabove.com/brand.md) ### Legal - [Privacy](https://www.andabove.com/privacy-policy.md) - [T&C](https://www.andabove.com/terms.md) ### Our partners - Anthropic - Google Cloud Partner - Vercel - Webflow Certified Partner ### Recognition - [Top Clutch.co Artificial Intelligence Company England 2026](https://clutch.co/profile/above-1?utm_source=clutch_top_company_badge&utm_medium=image_embed&badge=127900) ### Contact [Contact](https://www.andabove.com/contact.md) 1st Floor, The Binary, 27 Copperfield St London, SE1 0EN © 2026 &above™