Tom Roberts, Head of Digital Communications at Barratt Redrow, discusses how monitoring AI assistants reveals the sources shaping brand reputation, enables communicators to replace assumptions with actionable evidence, and shows why owned content, consumer platforms and forums may influence AI generated answers more than earned media.
1. What was your personal PR highlight of 2026?
2. What might surprise comms leaders in the next two years?
3. What is something you believe about AI in the communications industry that some communication professionals probably would disagree with?
4. What is the biggest benefit of using the Meltwater platform for your team?
1. What was your personal PR highlight of 2026?
My highlight this year was building something rather than launching something. Since the spring I have put together a monitoring programme that tracks how AI assistants describe our brands and our sector, running a large set of questions across all of the main models on a repeating cycle. It now sits on tens of thousands of answers.
I expected a scoreboard. Are we visible, are we described fairly, how do we compare, and so on. What we got instead was a list of things I could go and fix, not in a quarter and not in a strategy paper, but that week. Some of the citations and patterns were uncomfortable. Some of it was just strange. There are sources shaping what people are told about us that I would never have thought to look at, and would not have found by any other route.
Half the value has been in learning how these systems respond to different questions. Ask something broad and you can receive a diplomatic answer that reveals very little. Put two options side by side, require a choice and then test the pattern across a large dataset, and you begin to see distinctions that a single open question can obscure. Working out how to get the best results turned out to be as useful as the answers.
The bigger prize has been internal. Communications and Corporate Affairs teams spend a lot of time arguing from judgement. Turning up with a dataset nobody else in the business has, on a subject everyone is already curious about, changes that conversation. It has opened up conversations that were not happening before, and it means I have answers where I used to have a view.
For a fraction of the cost of a campaign, it has changed what I can see and what I can act on. I would not want to go back to assumptions.
2. What might surprise comms leaders in the next two years?
How quickly the habit sticks.
Most of the debate is about what the models can do. I think the surprise will be about what people do, and how fast. I notice it in myself. I talk to an assistant through the car on my drive to work now, in the slot where I used to put a podcast on. When I want to check something I have been told, my first move is no longer a search engine, it is to ask a different assistant and see whether the two agree. I see the same shift in colleagues, friends and family, and not one of them sat down and decided to change how they work. It happened on its own, once the tools got good enough.
What follows from that is harder than it looks. With search you could take the handful of queries that mattered and work on them. This is natural language. People can arrive from any angle, mid-conversation, sometimes by voice, and you cannot predict the question or tidy up the answer afterwards. You have to be in the material the answer is built from in the first place.
I am not suggesting anyone buys a house off the back of a chat with an assistant. But they will use one to weigh up an area, compare two builders, work out what to ask on a site visit, even talk through which plot on a development suits them best. That is the considered part of the decision, and it is now happening somewhere that until recently we weren’t able to see properly.
You will not spot it in your traffic either. Referral traffic from assistants is tiny, and I suspect it stays tiny, because when I am using these tools I do not want to leave them to visit a website. That is the whole appeal. So if you are waiting for your web analytics to tell you this is real, you may be waiting a long time.
3. What is something you believe about AI in the communications industry that some communication professionals probably would disagree with?
That AI is not the rescue of earned media, however comforting that idea may be to those of us who have built our careers in PR.
The line I keep reading is that because assistants read the press, PR matters more than ever, local titles matter more than ever, and the press release is back. I would certainly like that to be true. I have spent my career in and around PR and I am a huge believer in its value. Quality coverage still does what it has always done. It shapes what people think and it sticks in the mind, and will continue to have a vital role to play.
This framing is just not what shows up in the data, however. When I look at what actually feeds the answers people are given about our sector, earned coverage is not the engine. Our own pages are cited more often than the national, regional and trade press put together. Ahead of both sit consumer advice sites, review platforms and forums, along with a good deal of material nobody would ever think to monitor. Within our data there is a single post in a public Facebook group, for example, that has been cited as a source by five different assistants, across nearly a hundred different questions, on every single day we have sampled. Previously I would have had no way of knowing or tracking that.
Looking at this from another perspective does reveal better news. This is far more influenceable than most communicators assume. What you publish, how you word it, where you host it and how it is structured all affect whether it makes it into AI’s canon, by which I mean the pool of material these systems keep returning to when someone asks about you. I have watched our own wording travel into answers almost intact.
So the opportunity is not more announcements. It is treating what you already control as source material for machines as well as copy for people.
4. What is the biggest benefit of using the Meltwater platform for your team?
I have been a Meltwater client for the best part of eight years, first for media monitoring and then also for social listening. The platform I use now is a long way from the one I started on.
GenAI Lens is the part that changed what I could do. The value is volume and frequency: a large question set across ten models, refreshed on a regular cycle, and exportable so I can interrogate it properly. Scale is what turns an interesting idea into evidence, and evidence is what gets you a hearing internally.
The biggest benefit is not any single view, it is being able to ask what AI assistants are saying about our brands and then, in the same place, ask why. When something moves, I can look across at what the press has been doing, or a spike on social, or a Reddit thread gathering pace, and see whether they line up.
If I had to quantify it, it would not be hours saved. It would be that I now have answers to questions I could previously only have an opinion about.
About Tom Roberts
Tom Roberts is Head of Digital Communications at Barratt Redrow, with responsibility spanning digital communications strategy, corporate channels and online reputation. Before joining Redrow in 2017 as the first hire into its newly established communications team, he held marketing roles across the education, professional certification, sports and leisure sectors. Tom holds an MBA from Aston Business School, where his research examined how online reviews influence purchasing decisions in UK housebuilding. His current focus includes understanding how media, digital platforms and AI-generated answers shape corporate reputation and customer decision-making.
