Start with the problem: what two founders learnt about AI engagement

Written by Beverly Jones, Producer, AI Expert Podcast

In the first episode of the AI Expert Podcast, Charisma Mehta of Project Alix and Tiernan Mines of Hello Lamp Post shared the practical lessons they have learned on using AI to help public services listen and respond to the people they serve.

They both agreed there is no shortage of noise about AI in public services. What is harder to find is a clear account from people who have put it to work, with real residents, in real places. Their products are very different. Their advice for public sector leaders overlaps a great deal.

Charisma is a co-founder at Project Alix. It is a repairs intelligence and remote inspection platform used across social housing. A resident or operative takes a few photos and adds a few sentences about a problem, such as damp and mould. The platform asks follow-up questions in real time and produces a report for the surveyor, including a view on severity and likely next steps. Charisma said this has helped improve first-time fix rates and makes sure the right trade is sent from the start. If damage is near a plug socket, for example, an electrician goes out first.

Tiernan is a co-founder at Hello Lamp Post. It describes itself as an AI-powered digital team member for physical places. People ask questions by text or voice, and the platform uses their location and live data to answer. It can also take action, such as passing a request for assistance to the right team. Tiernan described how, at Glasgow Airport, the assisted travel team estimated that 80 to 90% of the queries reaching them were not about assisted travel. By his account, the platform cut those repetitive queries by 50% within the first couple of months.

 

Start with the pain point

Asked what genuinely works, Tiernan’s first answer was simple. Start with the problem or the outcome you want, not the technology. “It might not be AI that solves it,” he said.

His second point was about partners. A platform built to solve problems for everyone is probably solving shallow ones. He suggests looking for a partner who understands your sector and is willing to learn alongside you, because no supplier has seen every use case.

He also urged leaders to start small. When a room full of people starts listing use cases, it is tempting to take them all on at once. That rarely leads to a quick result. Pick one, prove it, and build from there.

 

Show, don’t tell

Charisma’s approach is to get people hands-on early. Her team often asks new clients to try to break the tool and tell them what they find. It helps people see where AI will help them and where its limits are, and gives them a sense of ownership.

She describes AI as working well on an 80-20 rule. Let it take on the common queries and issues, and keep human judgement for the 20% that needs it.

People can also surprise you. Charisma recalls a client whose senior team was nervous about giving AI to a group they did not see as technical. By the next session, two team members in their late fifties had come back with five new ideas and mock-ups, already tested with service users. Another workshop, at 2pm on a Tuesday in Swindon, drew a large turnout of people keen to have their use cases heard.

Her advice is to listen to the sceptics as well as the champions. People with reservations often understand things others can’t see.

 

Freeing up time, not cutting staff

Both our Podcast guests were clear that their products are not about reducing headcount. Hello Lamp Post has no app. It works through channels people already use, such as WhatsApp, SMS, QR codes and web chat, to keep barriers low. Tiernan accepts that no channel reaches everyone. Taking routine questions off staff, he argues, gives them more time for people who are digitally excluded.

 

Procurement, cost and the environment

The conversation turned to how organisations buy AI. Tiernan expects the environmental impact of AI tools to matter much more in vendor selection over the next few years, noting that the Environment Agency, one of Hello Lamp Post’s partners, measures the carbon reduction impact of using the tool. He also flagged that usage-based pricing can make long-term costs hard to predict.

Charisma suggests involving procurement teams early, and treating contracts as three-year partnerships you hope to extend, rather than buying a large platform with features that never get switched on.

 

What will age badly?

In the closing hot takes, Charisma pointed to the energy spent on meaningless AI work. Tiernan named two things: organisations putting company or personal data into general AI tools, and the belief that a generic large language model can solve thousands of specific problems.

Looking ahead, Tiernan is watching agentic systems that review and retrain themselves with less human oversight. Charisma is watching AI in the physical world, such as robotics and autonomous vehicles.

 

Watch the full episode

The full conversation covered much more, including the self-checking AI agent Project Alix uses to review every submission and Hello Lamp Post’s rollout across Los Angeles for waste and recycling.

Watch Episode 1 on YouTube, read the transcript and find the guests’ details on this page. https://aiexpert.digileaders.com/podcast/


Read More Collaboration

Comments are closed.