How a 187-year-old property firm is rebuilding around AI

Bidwells has set itself a public target: double revenue by 2030 without a proportional increase in headcount. AI is part of the plan for the 500 person property consultancy, but it’s not the plan on its own.

“AI alone won’t get us there,” says head of data and AI Ben Lee, the man tasked with understanding how AI can help Bidwells to this target. “Productivity gains are going to be incremental. But the real question is: how can we enable our people to do more so they can go out and grow the business?”

Ben started as a journalist, then spent two decades in real estate, joining Bidwells nine years ago as a comms manager. He’d been talking to the business about Generative AI for nine months before ChatGPT launched to the world. “When it did, people thought I could see the future,” he joked. 

Bidwells could have hired a machine learning expert – but as he put it, “we weren’t ready for that level of sophistication.” They needed someone who understood the operations, the data, and the people.

From AI council to innovation committee

Bidwells formed an AI group a couple of years ago, and Ben doesn’t think there was anything wrong with that. But his team has since developed into a broader technology and innovation team with a tech and Innovation committee that represents the wider business. “Quite often, AI is not the answer to the team’s problem. Let’s put the technology to one side for a second. What is your goal? What problem, if we remove it, will enable us to meet that stated goal of doubling the size of your business?”

The committee covers all ten of Bidwells’ departments. Before it existed, teams ran their own due diligence on technology. “You end up with duplication, or you end up with people not quite finding the right software solution and ending up with a system they’re not getting value from.” The cross-departmental view helped reduce that duplication. “A lot of people are all trying to solve the same problem. They’re the ones we then tend to gravitate and support.”

Where AI scales across the business and where it doesn’t

Real estate is a repeatable business in ways that aren’t always obvious – how meetings are run, how leases are reviewed, how scoping reports are written, or how site appraisals are carried out. Across all of Bidwells’ departments, Ben describes “a whole layer of things which are effectively document generation exercises with some orchestration of data.”

Bidwells prefers to build internally when they have the capacity. “But there comes a point where actually there’s a solution that fits our requirements very well. Let’s just go and buy it. Focus our in-house resources on something which we can’t find out there.” Ben calls it portfolio management – sort projects into buckets from self-serve through to enterprise-wide, and focus internal resources where a vendor solution can’t help.

But the work Bidwells’ clients value most sits above that layer – tailored advice built on specific market knowledge across Cambridge, Oxford and London’s Golden Triangle. That’s not highly repeatable, and it’s not where AI scales easily. Clients want support shaped to them, not a standardised output.

Ben’s team tested that with a hackathon. They took a real problem from Bidwells’ planning department – the kind of client-specific challenge that can’t be solved with a generic tool – and gave it to 40 university students with no preconceived ideas about how to approach it. They sent detailed briefing material ahead of the event so teams arrived with a solid understanding of what the business was trying to achieve. The students came from some of London’s top real estate programmes, in teams mixing machine learning, real estate and geography.

“The people I really listen to in all of this are the people that aren’t talking about technology,” Ben said. The hackathon illustrated why. The winning teams weren’t the most technically impressive – they asked the planners the most questions before touching any tools and spent as much time as possible understanding the problem before choosing their approach. They’re now coming in to do further work with Bidwells.

“It is quite easy sometimes to jump straight to, well let’s do this really spectacular machine learning exercise, when actually the winners didn’t really need to do that.”

Governance first, then tools

Bidwells tested multiple call analysis products that delivered great insight but were expensive. But the value was limited by whether the team had capacity to act on what it surfaced. “These tools are amazing. But they’re only as good as the humans on the other side making the decisions.”

That’s one kind of problem; the other is too many tools with no one managing the overlap. When Microsoft launched the Power Platform, its low-code platform for building applications, we saw what happens without clear governance and structure, Ben said. “You ended up with chaotic, inconsistent dashboards and disconnected solutions. No one quite knew where everything was. We may get that again as more and more companies build more and more AI Assistants, Agents, workflows, without clear governance.”

Now they’re more deliberate about it. There’s a trade-off between governance and speed – Bidwells handles sensitive client and property data that has to stay in the UK. “There’d be plenty of people we couldn’t work for if we didn’t do that.” So anything deployed at scale goes through the Microsoft stack. But they’ll prototype with third-party tools where they can move faster. A year ago, Microsoft’s tooling couldn’t match the speed of those platforms. That’s changed. “You can build a co-pilot now in a couple of hours in a way you couldn’t a year ago.”

People aren’t always moving as fast as the technology

Ben admits he’s surprised by the gap between how fast the technology has moved and how slowly people have followed. “I thought there’d be lots of people like me spending time learning, teaching themselves how to use the new platforms and tools, getting excited by the potential.” But “the pace of technology is still going super quickly and it’s way ahead of the way people are using it.”

He understands why. “They’re thinking about the job that they love and doing it in the way they always have.” But without direction from the top, that doesn’t change. “If you’re not clear about what technology means to your business, people don’t quite know how to engage with it.”

His advice to himself looking back: be patient. “It’s going to be a bit slow. That would have probably disappointed me at the time. But I think it’s been good in a lot of ways because it’s meant that you can be more strategic and more thoughtful about why we’re doing things.”

Every employee now has a performance objective – identify one area of their work that could be supported by AI. “It’s not a kind of, if you don’t do it, you’re fired situation, but it’s an explicit request that everybody in the business needs to think about how we can use technology to do our job better.”

“It could be as small as summarising a document by using Copilot. That’s just getting people thinking. It could be I’m going to build something which is going to deliver a new service for my clients. AI is not really an optional activity anymore.” Whether that builds real capability or just box-ticking is something every company making this move needs to watch.

“If you want to go and do this stuff, you have to believe it. You have to commit to it. It isn’t straightforward or easy.” But he also thinks the barriers to entry have never been lower – someone starting fresh today, with the right curiosity, can get up to speed quickly.