AI adoption depends on much more than the technology. Leadership, culture and organisational readiness determine what happens next.
That's where our third Built over Breakfast landed, on the topic of AI Beyond the Pilot.
Around the table were senior leaders from banking, consulting, utilities, technology, HR and founder-led businesses, each bringing a different perspective on the same challenge.
The conversation ranged from org design and governance to a problem I hadn't heard put quite so clearly before: AI can now produce work faster than organisations can absorb it.
Here are my main takeaways:
If businesses want AI adoption to stick, senior leadership has to take part.
That means CEOs and executive teams using the tools themselves, making time for AI and showing where it fits into real work. Sending everyone on a training course while leadership carries on as normal won't create the same change.
Getting adoption right across the organisation also means AI training shouldn't look the same for everyone.
Younger employees are generally more comfortable bringing AI into their everyday workflows. More experienced people have established processes that already work, and some are understandably reluctant to change them.
Someone who has spent 20 or 30 years adapting to new technology has a very different starting point from someone who used generative AI through university. The training, communication and involvement need to reflect that.
There was also a simple point about involving people early. If teams help shape how AI gets introduced, they have more reason to engage with it.
Most organisations still look like a pyramid: a broad junior base supporting a smaller number of senior leaders.
AI could flatten that into a diamond, with more responsibility sitting in the middle of the organisation.
As AI takes on routine execution, organisations may need fewer junior roles and more people focused on judgement, oversight and specialist expertise.
Longer term, that could become more of a rectangle, with responsibility and specialist capability spread more evenly across the organisation.
There is an obvious risk with that. Junior roles build future senior people.
If organisations remove too much entry-level work, they still need a way for people to build experience, judgement and domain knowledge. Otherwise, when senior people leave, they take a huge amount of knowledge with them.
AI can change where people start. Businesses still need to think about how they develop them.
If AI gives someone five hours back every week, what happens to those five hours?
One example discussed around the table involved automating more routine customer service work and retraining people into a more specialist design function.
The important part wasn't the automation. It was what happened to the capacity afterwards.
Those people learned new skills and moved into higher-value work.
That’s where return on time becomes useful. Saving time alone doesn't tell you much. You need to know where that time goes next.
So the ROI question shifts. Not what did the tool cost, but what did the reclaimed time produce.
Buying a product and looking for somewhere to use it is much harder than solving a business problem with the right technology.
AI is no longer limited to chatbots and copilots.
Some organisations are already running autonomous AI agents with names, defined functions and asynchronous workflows alongside product teams. Others are building AI personas around colleagues, using previous emails and responses to test how ideas, decisions and communications might land.
As AI takes on a bigger role inside organisations, governance, security, data access, audit trails and accountability become harder to separate.
Someone still needs to own those decisions.
One example shared over breakfast is this:
An engineer spent one week working intensively with AI. It then took the wider team roughly six weeks to absorb the work, review the outputs and decide what to do next.
AI accelerated execution. Decision-making didn't.
That changes the productivity conversation.
Generating more work is no longer the challenge. Organisations need enough experienced people to review it, apply it and turn it into action.
It also raises a longer-term question. If AI takes on more routine work, how do organisations continue building the next generation of experienced engineers and leaders?
AI can't fix weak foundations.
Strong data, connected systems and a clear understanding of the problem all need to come first. Without them, organisations risk building AI on top of fragmented data, legacy infrastructure and disconnected ways of working.
The same thinking applies to implementation.
Start with the problem and the outcome you want to achieve, then decide where AI adds value.
AI doesn't replace the need for good foundations. It depends on them.
Every Built over Breakfast ends the same way: guests leave one thought behind in the RemoBook.
This time we asked:
What will AI not have replaced in 12 months?
One guest, a leader at a global consultancy, wrote:
Judgement is still very much something humans have to get better at. AI will provide so many options, choices and possibilities and enable work to be accelerated, so having good judgement will become even more important.
AI makes work faster. Judgement decides what happens next.
That's why the organisations pulling ahead aren't simply adopting AI. They're building leaders and teams who know how to question it, apply it and make better decisions because of it.
Built over Breakfast started as a small executive roundtable. It's quickly becoming one of London's fastest-growing leadership networks for AI conversations.
From September, Built over Breakfast is moving to two sessions a month. We're also launching our first collaboration event, with more industry-focused breakfasts to follow.
The format stays the same: small, curated executive roundtables built around honest conversations and valuable connections.
If that sounds like your kind of room, drop Ethan Cohen a message to register your interest for an upcoming Built over Breakfast.