Why Just Move In rebuilt from the ground up

Watch the interview here

Just Move In helps people set up the essential services they need when they move house: energy, broadband, insurance, coordinated through partnerships with estate agents and mortgage brokers. With 10 years in business, they process around 20,000 moves a month and have earned a 4.9/5 on Trust Pilot from over 3,400 reviews. By any normal measure, the product was working.

But last year, they decided to rebuild it.

The result is Jay, an AI-driven platform that handles the full customer journey from move-in through to ongoing home management. Jay is a replacement for the old product, built on new foundations while the existing business kept running at full volume.

Last week we looked at why so many companies believe they’re further along with AI than they really are. This week, we dig into what it took for Ross to actually close that gap.

The old infrastructure couldn’t scale

The previous setup ran on a combination of digital tools and phone calls. It worked for the core services, but they could see a limit. If Just Move In wanted to go deeper (tailored financial products, property-specific recommendations, long-term relationships) the old infrastructure couldn’t support it. Not without hiring a large human team to manage every interaction, and that wasn’t efficient.

Ross wanted someone who comes back a month or a year after their move to not have to start from scratch. The platform should already know their property, the services they signed up to, and what they might need next.

Jay can see the property you’re buying (or renting) and its value relative to others on the road. Ross described what that enables: “Not just ‘here’s a loan,’ but ‘here’s a specifically tailored loan to improve the value of your property, and here’s a builder who’s available to start working on that.’ We’re helping the customer get to the end decision faster than if it was just them on their own.”

Their Series A closed at Christmas 2024, as those decisions were being made. Ross described the tension well. “There’s this balance you have to strike between delivering against the plan you put in place for your current business, and making it clear that you’re seeing the direction of travel and operating to that new bar. You have to do both.”

Making it work inside the company

When Just Move In announced the rebuild, they brought in new designers, product experts and engineers to get Jay off the starting blocks. But the harder challenge was shifting the culture of the existing team.

People with 10 or 15 years of industry experience had seen their worlds change rapidly. The Just Move In leadership team  needed them thinking AI first, not treating it as something to check their work against after the fact. “AI now builds the code, the humans then review it.” That inverts how most experienced teams operate.

“We are by no means the finished article in terms of everybody across the entire organisation using AI naturally first every single day.” But new joiners coming in with that mindset are helping the rest of the organisation get on the same page. And what seems to have helped accelerate things most is leading from the front.

Ross, who is not an engineer, recently built a fully functioning product feature for the business. A proper website with front end, back end, database, payments and admin area, all using AI tools, no product team, no engineering sprint, and it’s already live.

“That wouldn’t have been possible a year ago. Even six months ago, I could get so far, but then it would fall over when it came to the payments and the database components.” That used to require a request to be spec’d out, resourced, and maybe shipped in a year. Ross went from idea to working product on his own.

“Don’t be afraid of it. Lean into it, make mistakes, share in public.” They’ve started encouraging the same across the team, with Slack channels where people share what they’re experimenting with. No formal programme, just a culture of trying things and being open about what breaks. Getting good at prompting is becoming a core skill.

A staged rollout rather than a big bang

Jay went live in mid-December, but they deliberately held back.

Initially the AI just had conversations, no commercial services or selling. Just helping people with their moves, answering questions, building rapport. They tested on their own direct customers first for a couple of months before turning it on for partner channels. They work with hundreds of partners, so it had to work before wider release.

“And of course it was breaking a lot,” Ross admitted. “There are going to be a lot of bugs when these new tools roll out. We’re something that is completely new. Nobody has done this before, specifically in our space.”

Every conversation Jay has gets reviewed. Their own AI evaluates interaction quality, then humans review on top of that. When the AI gets something wrong, they feed it back in. The top metric is conversion to actually using a service, not conversation volume or time saved.

The staged approach also surfaced something they hadn’t expected. International expansion was already on the roadmap, but while reviewing Jay’s conversation logs, the Just Move In team noticed that international customers, roughly 10 to 20% of their base, were interacting in their native languages. “If your first interaction is in your mother tongue, that’s an incredible touchpoint,” he said. “You’re naturally going to lean towards that product for whatever else you might need after the move.” 

European expansion had always been a possibility, but the rebuild had made it far more accessible than they’d assumed. They already have global partners and a supplier base covering multiple countries, and the product can operate in any market without hiring a local sales team. They’re now planning to enter a new market next year, testing on a small trial basis before committing serious resource.

What to build and what to buy

Just Move In built the core Jay platform and the customer experience around it themselves, that’s proprietary. But in the back office, they’ve taken a more pragmatic approach, buying tools where it made sense while accepting they might replace them later.

They invested in a call analysis product that was expensive and delivered great insight, 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.” Ross admitted they’d fallen over in certain areas on exactly that point.

We hear this regularly from companies working with aibl. The tooling decision is rarely the hard part, it’s having the right people focused on making the tool deliver.

His advice for vendor selection: run a proper RFP, review four or five suppliers against clear criteria, pilot before committing, and make sure you have the right people internally to deliver on it. “You can end up with lots of software with no one using it.”

The tooling landscape is shifting quickly. Platforms they already pay for are releasing features that will likely replicate what they’re currently buying separately. Voice AI quality is improving so fast that the gap between building and buying closes all the time. For now it’s about balancing the two, and staying ready to switch when things move.

Nobody has it figured out

Ross was clear on what it takes. “If you want to go and do this stuff, you have to believe it. You have to commit to it. It isn’t straightforward. It isn’t easy.” But he also thinks the window is wider than most people assume. The landscape moves so fast that someone starting fresh today could be at the front of the pack within a month.

“We’re still figuring it out ourselves as we go. We don’t have all the answers. I don’t think anybody does. I think that’s the thing that will give people a lot of comfort.”