When Habs Kim’s family moved from South Korea to the UK, one of their first encounters with the British state was one of the most feared entities on the British Isles: The planning department.
Her parents were trying to open a takeaway shop. All they needed was permission for a change of use of an existing location. It should, in theory, have taken eight weeks. It took six months.
“I found that was incredibly impactful on me as a child,” Kim said, remembering the stress it put on her family. “The impacts that planning delays actually have on very ordinary people and ordinary families have always stuck with me.”
That experience now sits behind Xylo, the UK govtech startup Kim cofounded with former lawyer and operator Dermot O’Riordan, which has now raised £2.8m in pre-seed funding to help local authority planning officers process applications faster and more accurately.
The round was led by CapitalT, with participation from Common Magic, Sure Valley Ventures, Tiny VC and Endurance Ventures, plus angels from DeepMind, Gensyn, Dawn Capital, Mark Ransford and exited govtech founders.
Xylo is building AI agents for planning teams. The pitch is not that AI should make planning decisions, but that AI agents can rescue planning officers from drowning in admin, and give them what they need to make a decision about an application.
“Planning officers historically reviewed everything on paper,” said Kim. “They had a secretary called ‘Elaine’. Elaine would handle all of their admin. And then in the 2000s, the software arrived, Elaine got cut from the budget.”
The result, Kim says, is that officers now spend “half their day” on admin.
“So our officers tell us that Xylo is like having ‘Elaine’ back,” she said. “Our agents do the legwork. They hand over the first draft report in minutes, and the officer edits the report and then makes the determination.”
The company says it was built after the founders spent more than 300 hours interviewing planning officers, including work with Leeds City Council as a design partner. Xylo claims its platform helps councils process 40% more applications per officer per month, while 83% of officers say it helps them work more accurately.
Kim admits “we have absolutely no business doing government,” but they are “just two ordinary citizens who thought this has to be a problem that we could fix.”
That outsider status has, perhaps, also helped shape the product, after they spent months sitting beside the offices, shadowing them and trying to understand why previous software had failed.
That process led Xylo to focus on the unglamorous work that happens before a planning officer can make a decision: assembling files, researching policy, reviewing site context, summarising consultations and preparing the first version of a report.
However, Xylo only goes so far.
“The platform stops doing its work before the decision is actually made,” said Kim. “We help with everything up to decision. And we leave it up to the officer to apply their judgment, do the review and actually make the final decision.”
She added: “We believe very strongly that the officer is the expert in this case and they need to be the person to actually approve and make the final decision.”
Given that planning is a legal process, AI tools in public decision-making are likely to face intense scrutiny. Kim says Xylo uses a mix of major AI models, but wraps them with evaluation, regression testing and comparisons against final official-published reports.
The company is initially focused on the UK’s busiest local authorities, where planning teams are already under pressure. It is also looking at the US ‘permitting’ market, where Kim says the same admin burden exists despite a different departmental structure.
