What we learned being judged on AI Innovation

Ten minutes on the clock. A football box suite pressed into service as an interview theatre. We were there because Door4 had been shortlisted for the Red Rose Awards in the AI and Tech Innovation category. Proud moment. Also, mildly terrifying.

We went in ready to talk about models, stacks and shiny demos. We came out talking about operational change, proof of value and whether our team can explain our approach in plain English. The judges did not ask what we expected. That turned out to be the most useful thing about the whole process.

What the judges really cared about
Yes, they wanted to see innovation. But they kept steering us back to the same place: evidence that it works, for us and for clients. Not just can you build it, but does it move the needle, and can you show how?

Leon Calverley, Founder, Door4: "We are accustomed to being interviewed. The questions were reassuringly focused on the commercial angle, as well as the technology. We base our innovation on People, Process, Technology and the judges’ commercial questioning matched this."

In other words, we have sailed past the early hype. This is the ‘now what’ phase. That tracks with the wider market. McKinsey estimates that generative AI could add trillions in value to the global economy, but only when it is embedded in workflows with measurable outcomes. The potential is not in the model, it is in the process you change and the decision you speed up.

There was also a healthy scepticism about theatre. The judges pressed for clear links from idea to pilot to impact.

What we showed them, in practice

We kept it practical. Rather than talk about what AI could do, we focused on things it already does for our clients and us.

Leon Calverley added: "We focused most, in the time we had, on two core elements of built technology. Our content synthesis workflows, including tone of voice matching. And our internal operating system, where agentic and workflow AI services hold the team to targets, surface headline data and flag coverage issues." 

That inside-out view mattered. If you do not use it yourself, why should anyone trust you to advise them. We also talked about client-facing tools that save time or drive revenue.

Sean Dwyer, managing director, Door4: "It was tough, with only ten minutes, to get across our multi-faceted approach.

"We improve internal efficiencies, improve client output for core services, deliver workshops that help customers map people and process to tech, and we build lightweight systems to automate work.

"My preferred example is our best-in-class content enrichment tool. It improves content quality, creates new business opportunities for us and saves clients time, all in one."

Culture came up too. Tools are nothing without people who know how and when to use them.

Darren Taylor, UX Lead, Door4 said: "We showcased our evolution well. What we could have emphasised more was Door4’s culture around AI, how we empower staff to embrace it, make tools available, and actively encourage learning and experimentation.

"Walking the judges through practical processes like the Instagroup warranty tool or Seddon workshops might have resonated more than internal tools alone."

We also touched briefly on other stacks that support growth: imagery generation embedded in creative workflows, lead sourcing with ideal customer profile scoring, and an analytics agent stack that turns data into next best actions for marketing teams.

Finally, the advisory side. Our Art of the Possible sessions are designed to surface the ‘we-did-not-know-we-could-do-that’ opportunities that sit between departments. In the interview, the judges were keen to understand how those conversations become real projects. The short answer: start small, instrument well, scale only when the numbers hold up.

The hardest question

Impact is easy to hint at and hard to prove in a ten-minute slot. The judges went straight for the specifics.

Sean Dwyer said: "I thought the focus would be on commercial impact. That was still the hardest thing to convey without pulling out our P&L and walking through time saved, clients won, clients retained and deals sold." 

That is the right question. If you cannot show a line of sight from a workflow or agent to revenue protection, growth or risk reduction, you are not done yet. Fair challenge accepted.

What we would do differently

We will be honest. We crammed. The deck was a touch text-heavy. The room setup made eye contact tricky, and team flow even trickier. But that’s what these pressure situations are about.

Darren Taylor said: "The host was impressive. The suite was an odd shape, more corridor than meeting room, which made team interaction hard. Only two judges dominated the conversation. The final slide asking Did Leon finish on time? Yes or No was a memorable talking point, though maybe not the lasting impression we intended." 

Leon would also have liked more time to separate our advisory and delivery work with crystal clarity. That is on us. As Leon put it afterwards, ‘enforced introspection is a gift’. We left with a clearer internal brief.

Three takeaways

• Start with the problem, not the model. Define the decision or process you want to change, the user journey it touches and the metric that proves it worked. No metric, no project.
• Pilot small, measure properly, scale only when it works. Instrument time saved, error rates and conversion deltas on the pilot. If the lift is real and repeatable, bake it into process and training. If not, kill it fast and move on.
• Your team must understand it, not just use it. Pair tools with enablement. Teach prompt patterns, data hygiene and when to switch the AI off. Confidence comes from practice, not policy PDFs.

Whether we win or not, the exercise forced clarity. If a busy marketing director cannot explain their AI strategy to their board in three sentences, it is not a strategy yet. We will keep doing the work that makes those three sentences easier to say. If you want a structured place to start, our Art of the Possible sessions and practitioner workshops are designed for exactly this moment. No hype. Just problems, process and proof.

Another closing thought:
Leon needs to learn how to keep to time.