.jpg)
We spent Day 2 of Engage Employee Summit, held in Battersea Park, London, immersed in conversations about the future of internal communications and employee engagement, and came away with one clear conviction: the organisations that will thrive are those that use AI to amplify human connection, not replace it.
Across seven talks we attended, from futurists and L&D specialists to comms leaders at the BBC, Knight Frank, Odeon and BDO, five themes stood out for us. We've pulled them together here, along with the insights and examples that resonated most strongly with us.
1. AI is a tool, not a threat, but trust is everything
The phrase "AI as a tool, not a threat" was practically the conference motto, and for good reason. Used well, AI can help comms teams synthesise thousands of data points, surface emerging themes, and free up time for the genuinely human work. But used carelessly, it actively destroys trust.
Edelman’s 2026 Trust Barometer data presented on the day showed that trust in senior leaders is already falling, with fewer than 50% of employees trusting their leaders in the most recent report, down nearly 10 points from 2024. We are 65% more likely to follow a leader whose values align with ours. That's an enormous responsibility, and one that AI generated comms can inadvertently undermine. (https://www.edelman.com/trust/2026/trust-barometer)
The Aldi example was striking: after the supermarket started using AI generated images for social media, impressions dropped significantly. Audiences noticed. They always do.
The consensus was clear: be transparent about when and how AI is used. The BBC has made this a non-negotiable, nothing goes out without thorough human judgement, and teams are trained to be open about AI's role in the drafting process. Transparency doesn't undermine credibility. It builds it.
2. Human voice is irreplaceable, and it's under threat
Multiple speakers sounded the alarm about what one panellist called "AI slop", content that looks fine on the surface but has lost the texture and authenticity that makes people actually feel seen and heard. The risk isn't just reputational. It goes to the heart of the employee relationship.
As one panellist put it plainly: "There is no replacement for human vulnerability."
Flutter described how remote and hybrid working had already created a creeping dehumanisation of comms, and AI risks accelerating this. They've adopted a "human first" rule: if a communication is substantive, personal, or related to change, it must come from a real person. They also noted a telling irony: employees are now deliberately dumbing down their writing to prove it wasn't written by AI.
The comms professional's role, one panellist argued, is increasingly to be the person who takes a colleague's AI assisted draft and puts the humanity back into it. That's not a criticism of AI. It's a recognition that the human layer remains essential.
3. “Internal comms is the infrastructure of trust”
This was Nazia Nathu's framing, Head of Comms at BDO and on the board of the IOIC, and it's one we'll be using a lot going forward. "If people don't trust the comms, they won't trust the change." It sounds simple. It's harder than it looks.
Nazia's session explored how trust has shifted over the last decade and a half. With teams more distributed, the meaning behind what you communicate matters far more than the mechanics of how you deliver it. The question every comms team should be asking isn't just "what channel should this go on?", it's "what do we want people to hear, feel and do?"
She made a powerful point about silence: leaders often wait for the "right moment" to communicate. But that gap, that silence, is deadly. People fill it with rumour and anxiety. Even when the answer is "we don't know yet," saying so is better than saying nothing. Treat people as grown ups. Tell them why.
One statistic from the IOIC Internal Comms Index landed hard: 1 in 5 employees don't feel connected to their team. In a hybrid and remote world, AI generated communications risk making this worse, not better. The antidote is honesty, regularity, and closing the feedback loop, even when the answer is uncomfortable.
4. Employee voice must move from survey to real-time listening
The panel on employee voice was one of the most practically useful sessions of the day. The shift being described isn't incremental. It's a fundamentally different relationship with data and listening.
Siobhan Heffron from Knight Frank described how they moved away from annual HR led surveys that took months to yield results. Using Workday Peakon, they launched a campaign called "Let's be Frank", manager owned and always on, giving leaders visibility of "people health" the same way they track sales metrics. The result? Leaders can identify challenges, track progress, and act, not just report.
Caroline Mogg made a point that reframed how we think about measurement: moving from eNPS to favourability scoring. eNPS tells you how someone feels. Favourability scoring starts to get at why, and that's where the real insight lies.
Odeon's approach was equally instructive. With thousands of staff who don't have work emails or mobile phones, their people team took iPads into canteens and communal areas to meet employees where they are. The principle generalises: if you're waiting for people to come to you, you're already missing the conversation.
The broader message: if you ask regularly, you must act regularly, and communicate what you've done as a result. Listening without closing the loop erodes trust faster than not asking in the first place.
5. The future of L&D is capability, not content
Kristen Budd from Synthesia delivered perhaps the most provocative session of the day, a challenge to almost everything we assume about how learning and development works. We have more training content than ever before. We are not better for it.
The stat that stopped the room: 90% of training content is forgotten within a week of completion. We've been measuring the wrong thing, outputs like lessons completed and modules finished, rather than capability, which is whether this person can actually do the thing under real pressure.
Her definition of capability was precise and worth holding onto: "the ability to perform a target behaviour, in context, under pressure, within a job aid." That's a very different goal from "completed the e-learning module."
Drawing on Bandura and Dweck, she outlined a four stage model that separates a course from genuine behaviour change: Inform, Demonstrate, Practise, Iterate. The crucial shift is in stage three. AI now makes it possible to create simulated practice environments where people apply skills in realistic scenarios and receive specific, direct feedback. The technology is here. The question is whether L&D teams are willing to redesign around it.
Her practical challenge to the room: take one existing lesson and rebuild it using this framework. Start with the capability you want to build, not the content you want to deliver.
What does this mean for us?
Taken together, these five themes point in a consistent direction: the organisations that will navigate this period well are those that double down on honesty, human connection, and the courage to communicate even when the answers are incomplete.
AI is genuinely useful, for synthesising data, drafting first passes, identifying patterns in employee feedback, and creating practice environments for skill building. But it requires human judgement, human vulnerability, and human accountability at every step.
The leaders who will build the most trust aren't the ones who have all the answers. They're the ones who lean into uncertainty with transparency, stay connected to what's actually happening on the ground, and treat their people as the capable adults they are. That's not a new idea. But it has never mattered more than it does right now.
Disclaimer: We used AI to transcribe our notes and then to write the first draft of this blog! But we also added our human touch and edited to better reflect our experience of the Summit.