CityFibre uses visual AI to quality check every fibre installation, up from a 20% sample. Installers photograph their work, and a model on the phone checks it, even with no signal. David Tomalin, Group Chief Technology Officer, told the story at aiblLIVE London 2026.
The first use case paid back in four months. The approach then carried CityFibre from 10,000 to a million homes connected a year, with no extra headcount. Here is how it worked.
All figures as stated on stage by David Tomalin.
Installers photograph each cable connection on an ordinary phone, and computer vision inspects the photo. Manual review could not cope with hundreds of photos at once, so sampling was the norm. With visual AI, CityFibre moved from checking 20% of work to 100%, and the first model needed about 1,000 training photographs.
Because many sites have no signal. Most data centres have no mobile coverage, and one acted like a Faraday cage. Storing photos to send later only led to revisits. About five years ago, CityFibre built visual inference into standard phones, so checks run on site without Wi-Fi or internet access.
If we’ve installed it clean in the first place, it should break less often.
David Tomalin, Group Chief Technology Officer, CityFibre
It made quality pay. Installers who miss the standard must revisit at their own cost, while proven work gets paid faster. Tomalin says that win-win began to shift construction industry culture. Early teams became advocates, which let CityFibre extend the approach into adjacent workflows and across its partners.
A full network upgrade with a tiny team. CityFibre made 55,000 network touches in 12 months with fewer than 10 people, retesting every fibre, and now sells symmetrical 8 gig services. The photo records also show what is installed, so agents can add business and service insight on top.
Scaling the model is about cultural adoption. It’s about helping people find the right way to work with it.
David Tomalin, Group Chief Technology Officer, CityFibre
David Tomalin is Group Chief Technology Officer at CityFibre, the alternative fibre network that has built over 5 million premises across the UK. He has spent most of his career in telecoms. He spoke on the AI Infrastructure track at aiblLIVE London on 20 October 2026.
If quality checks are slowing down your field teams, this is the model to borrow. The full PDF goes deeper than this page, in the speaker’s own words.
For CTOs, COOs and operations leaders running field, construction or infrastructure work. Six pages, about an eight-minute read.
