Public sector comms teams are already stretched. I sat down with our Chief Technology Officer, Hardeep Johal, and our Head of Value Delivery, Rosie Knight, to talk about where AI adoption genuinely helps, where it doesn’t, and why the human element still matters.

This wasn’t a polished recording. Blurry webcams, questionable lighting, and Hardeep protesting while I tried to take a photo of us. And looking back on what we ended up discussing, the irony wasn’t lost on me: there we were, three people fumbling with webcams while preparing to talk about the future of tech.

Once we were vaguely framed on camera, I jumped into the thing I hear most from teams right now: they’re overworked, overloaded, and expected to do more with less. And somewhere along the way, AI has been positioned as the miracle fix – which it isn’t.

Sarah: Teams are under pressure like never before – more channels, tighter resources, higher expectations. Where do you think AI genuinely helps?

Hardeep: Right now, it’s the repetitive stuff. Drafting content, generating variations, suggesting formats for different channels – it saves you from staring at a blank page. But I think comms teams are under-using AI on the analysis side: spotting trends, understanding sentiment, surfacing insights. Everyone thinks of GenAI as “the writing thing,” not “the thinking partner.”

As he said this, it struck me how rarely teams have time to dig into learning from their content. Everyone’s drowning in posting, not pausing. AI has the potential to change that.

Rosie: And I think we get stuck talking about efficiency. You know, the whole “It’s faster!” thing. Sure – it is faster. But the real value is accessibility. AI should make good comms more achievable, not just more frequent. Because if all it does is speed things up, you just end up stuck in the same cycle of churn.

What we’ve done with our Voice of the Community feature is a good example of this. We’re pulling together huge data sets that would’ve been completely untapped for comms teams before. And that’s the point – we should be using AI to open up space for the work that actually moves people.

Sarah: So more time back for teams, not more pressure to produce twice as much?

Rosie: Exactly. Less output, better quality.

“Speed won’t move an audience – depth will.”

That line stuck with me. Teams don’t need to make more noise. They need tools that help them think clearer.

Sarah: So if AI helps with repetitive tasks, where shouldn’t it be used?

Hardeep: Anything nuanced. Strategy. Crisis comms. Sensitive topics. Anything requiring real-time judgement or awareness of unfolding context. AI doesn’t know what’s happening outside the words you give it. And it shouldn’t be deciding that for you.

I found myself nodding, because I’ve seen AI drafts that look right at first glance but feel… off. Missing empathy. Missing context. Missing the sense of responsibility.

Rosie: And accountability still sits with the human. AI can suggest things, but you’re still the one who sends it. Especially when it comes to misinformation – there are tools like Grok scanning the internet and responding automatically, and if you’re not paying attention, you’re basically handing over responsibility without any real oversight. And it’s not even that easy to miss anymore. Our social listening and insights tool can alert you to this kind of information, and even summarise it for you. We’ve seen comms teams where Grok comes up as the top contributor, which is really risky.

And this is even more true for public sector teams across the UK and US. They’re not dealing with “Where’s my order?” or “I forgot my password.” They’re dealing with vulnerable people, safeguarding concerns, crisis reports and citizens who need real support – fast.

That’s the kind of work you don’t hand off to a bot. You need someone who can read between the lines, recognise distress, and respond with care.

Sarah: We say “work smarter” constantly. What does that look like in reality?

Rosie: Doing less. Not in a lazy way – just in a focused way. When you’re overwhelmed, something’s broken. AI should clear the noise so you can prioritise the meaningful work. The work that actually connects with humans. Most of the real thinking happens in what I call the “messy middle” – the part between the first idea and the final message. It’s where everything is half-formed, where you’re figuring out what you actually mean, where the emotion and nuance get shaped. AI can help tidy and clarify that stage, but it can’t replace sitting in it.

“AI can tidy the messy middle, but humans give it meaning.”

I loved that phrase – the messy middle – because every comms person knows exactly what it feels like.

Hardeep: It’s letting AI do the first pass. And helping teams interpret data, not just collect it. Anyone can generate reports. The value is in understanding what they mean and what you should do next.

We discussed how teams mistake “more” for “better.” Most public-sector comms teams don’t need more outputs – they need more clarity.

“A good brief is a good prompt.”

Rosie references how public-sector teams already know how to brief agencies. They know how to communicate tone, context, nuance. Prompting is just a new form of that existing skill.

Sarah: I think people worry they’ll lose their voice if they lean on AI too heavily.

Hardeep: And they’re right. You can tell when something is straight out of ChatGPT. It’s overly polished. It’s generic. And when teams do that, they lose credibility. I see stuff on LinkedIn and think, “come on, that’s not your voice.” Trust disappears fast.

Rosie: The funniest thing I’ve seen recently was a LinkedIn post that was literally one sentence – and it made me laugh, because you could just tell a human wrote it. No emoji bullets, no dramatic build-up, no over-polished structure. Just… a person being a person.

AI, on the other hand, has this rhythm to it – you can spot it a mile off. It loves a three-point structure, it loves a tidy arc. Sometimes that’s actually helpful because it cuts out the waffle. But when everything starts following that same pattern, it all begins to feel the same.

Sarah: Everyone says they want to sound authentic. What does that actually mean right now?

Hardeep: Two things: genuine empathy and honesty. Not a performance of empathy – actual care. And honesty about who you are and how you communicate with your community.

Rosie: And consistency. If you only show up occasionally or only during a crisis, trust won’t build. You can use AI to support consistency, but the tone – the heart of it – still needs you.

It’s true. Authenticity gets thrown around so lightly, but when you break it down, it’s very simple: show you care, and show up honestly. You can’t automate that. 

We drifted naturally into product – what we already do at Orlo, and what we’d love to do in future.

Sarah: How do we build tools that help teams stay authentic?

Hardeep: It’s about giving the AI the right foundations – brand guidelines, tone of voice, examples of past content, regulatory notes. That’s how you let AI support without replacing. And to be transparent – we’re not fully there yet. But the invisible AI, the stuff you don’t notice working? Sentiment analysis, identifying misinformation, highlighting high-priority messages, protecting staff with sensitive content detection – that’s where we’re strongest today.

I love this part because it captures how AI can do powerful things without shouting about itself. Not everything needs a shiny “powered by AI” sticker. Sometimes the best AI is the AI you don’t even see.

Sarah: Is there such a thing as too much automation?

Hardeep: Yes. Bots replying directly to citizens on behalf of humans? We’re not there yet. It’s fine to recommend responses – that’s helpful. But full automation of messaging, especially in public sector, is risky.

Rosie: It’s about weighting. Some things can be automated – simple FAQs, routine updates. But high-risk, emotional, or sensitive contact should always go to a human. There has to be a boundary.

And honestly, I think this is something public-sector teams already intuitively understand. They’re cautious, and rightly so.

Sarah: People worry about becoming replaceable. What do you think the future looks like?

Rosie: More time with humans. More face-to-face, more real connection. If AI is doing the repetitive work, you have more capacity to go out and meet the people you serve.

Hardeep: And it’ll highlight your value. Think of a sculptor: apprentices can chunk the marble, but only the sculptor adds the detail that makes it art. AI is the apprentice. Humans will always do the last 20% that matters most.

“AI can draft the first 80%, but humans deliver the meaning.”

I love that. It summarises everything.

Sarah: What excites you about AI, and what scares you?

Hardeep: It’s amazing how much it takes off my plate. Things that used to take hours take seconds. But what scares me? People getting complacent. Losing skills. Letting their curiosity fade. That’s when we get into trouble.

Rosie: Same for me. It’s exciting, but it still relies on us being thoughtful. Creative. Curious.

I think that’s the heart of it: AI strengthens whatever you put into it – but only if you’re deliberate about how you use it.

Sarah: One thing AI does brilliantly?

Rosie: The messy middle. You start it, AI tidies it up, you finish it.

Hardeep: Refinement. Polishing.

Sarah: One thing only humans can do?

Hardeep: Make the moral call. Feel. Be accountable.

Rosie: Be human, basically.

Sarah: One task you’d happily hand to a bot?

Hardeep: Drafting anything.

Rosie: Holiday planning! And temperature-checking ideas.

Sarah: And one word to describe the future of social?

Rosie: Trust.

Hardeep: Human.

By the end, it felt like we’d lived the messy middle ourselves – working our way through that in-between stage where thoughts and ideas slowly find their shape. And right at the end of all that, Hardeep said the word that tied it all together:

“Human.”

That’s the point of all of this.

AI in communication isn’t here to replace us – we can integrate AI into our strategies. It’s here to give us back the space to be more human in the work that matters.

More room for empathy.

More room for intentionality.

More room for the work that makes a difference.

And maybe… even more room for another cup of tea!