London Tech Week 2026 AI Insights: Why Data Quality Matters

Why the smartest people in the room weren’t talking about smarter AI — they were talking about cleaner CRMs.

London Tech Week – Around 30,000 people from over 130 countries, six stages, a royal keynote, and approximately one thousand uses of the word “agentic.”

Faye sent exactly one delegate. Just me — Liz with Conference pass, and my laptop taking notes as I tried to join as many conferences as I could. More on that later — let’s get to the actually useful part first.

AI was, unsurprisingly, the word of the week. It showed up in keynotes, on banners, in the coffee queue. But the longer I sat in sessions, the clearer one quieter thread became: nobody serious in the CRM world on the stages was saying AI fixes a messy CRM. They were saying it depends on a clean one.

Your CRM just became the front door, not the back office

Over on the AI Arena stage, HubSpot’s CMO Kipp Bodnar led a session on agentic commerce, walking through how AI agents are starting to do the browsing, comparing, and shortlisting that used to be a human buyer’s job. HubSpot and Dentsu made a related point to packed rooms: AI assistants are becoming the new “front door” to a brand, with a growing share of searches now ending without a single click through to a website.

That should give every CRM user pause. If an agent is going to read your account records, your deal stages, and your contact history, and act on them on your behalf, then the half-filled fields and duplicate contacts aren’t just an internal annoyance anymore — they’re literally the raw material the agent is reasoning from. Good CRM hygiene used to be a nice-to-have for the ops team. It’s now the thing the AI is actually running on.

Even the feel-good story was a data story

One of the more memorable moments of the week had nothing to do with sales tech: the Prince of Wales made his first London Tech Week appearance, chairing a panel on using data to prevent homelessness. His Homewards initiative used the week to launch the UK’s first Homelessness Data Lab with Salesforce and LandAid, bringing together more than 25 organisations to look for the early warning signs, rent arrears, benefit changes, family breakdown, that tend to show up in data long before someone loses their home.

It’s a genuinely good initiative, and also a tidy real-world example of the same principle from a very different angle: the “AI” part of that project is really just well-structured, well-governed data, organised so the patterns become visible. Take away the clean data, and there’s no early warning system left to build.

What this means for us, and for the clients we talk to

This is the pitch we should be carrying back to clients: AI isn’t a replacement for good CRM practice, it’s the reason that practice finally matters enough to fix. A few basics worth repeating in every onboarding and QBR:

Standardize the fields that matter (lifecycle stage, owner, source) so an AI tool isn’t guessing at what “done” or “qualified” means.

Run deduplication on a schedule, not just when something breaks.

Give every record a clear owner and clean permissions, so there’s accountability for the data an agent will eventually act on.

Use a propose-then-approve pattern for AI-driven cleanup: let the tool suggest changes, but have a human confirm before anything writes back to the CRM.

None of this is exciting. That’s rather the point — and exactly the conversation London Tech Week kept circling back to, session after session.

By Elizabeth Agudelo, Consultant

Elizabeth Agudelo is a CRM Business Consultant and Project Manager at Faye, where she leads CRM transformation initiatives at the intersection of customer experience, data, and enterprise technology. With over five years of experience in CRM consulting and delivery, she partners with organizations to translate business requirements into scalable, high-performing CRM solutions built on platforms such as SugarCRM.