How to use AI to listen, understand and act without losing the human touch

With tech leaders arguing in the news again this week over whether AI development should slow down for safety checks, the ongoing debate can make this whole topic feel unpredictable. Yet behind the headlines, we all know AI itself is nothing new. It has been sitting in software platforms, digital channels, and news feeds for a long time, but we’re still figuring out how to actually use it. And we know that in public sector comms, it remains a tricky conversation.

Public sector comms carries real weight. You are managing public trust, handling sensitive community issues, and supporting residents when they need help. Taking a cautious approach to AI makes sense when the stakes are that high.

To help alleviate some of that pressure and to address some of the challenges, myself and our Head of Customer Success, hosted a webinar. Our message was simple: AI needs to work as an enabler rather than a replacement.

Here is a look at what we discussed, and how public sector teams can use AI optimised tools, like Orlo, to handle routine admin, freeing up time for the work that requires real human empathy and judgement.

Recent research shows that 74% of public servants use AI tools at work, with most adopting them over the last year. At the same time, 9 in 10 UK public sector workers say their organisation is not properly set up to get the full value out of the technology.

Teams are constantly asked to stretch budgets while community expectations keep rising, and while tech companies debate the future speed of development, frontline teams still have daily messages to answer and issues to resolve. Stanford Professor Fei-Fei Li put it well when she said that “AI is about amplifying human potential, not replacing it”.

The main job for public sector leaders today is making sure these tools are safe, secure, and genuinely helpful for staff.

Hesitation around AI usually comes down to control. No team wants an unsupervised tool publishing the wrong thing or sending private resident data off to public training models.

Orlo is built specifically for public accountability. We stick to a plain rule: human-led, AI-optimised. Here’s how it works:

• You keep final say: Every draft, summary, or suggested response is just a suggestion. A real person reviews and approves every action before anything gets posted or sent.
• Specific, practical tasks: Orlo’s AI handles repetitive inbox sorting, text editing, and quick summaries. It does not touch policy decisions or professional judgement.
• Your data stays private: Community conversations are processed only to support your service. They are never used to train public AI models.
• Sits inside your current setup: The tools live directly inside the inbox, composer, and reporting screens your team already uses every day.
• Government-aligned security: Over 400 public sector bodies trust Orlo because we maintain the strict security and compliance standards required for public work, including two-factor authentication, single sign-on, role-based access control, account takeover protection and ISO, SOC2 and Cyber Essentials Plus. 

We have built features directly into our social media management, social and media  listening, and Voice of the Community solution, to shave hours off routine admin:

AI Content Studio

• Post creation: Generative AI will draft social post copy from a prompt directly in the composer, leaving your team to tweak the tone and add local details.
• Rewording: Rewrites approved copy for tone, length or channel without changing the core message, so social media algorithms do not penalise your account for posting identical text.
• Hashtags: Suggest contextually relevant hashtags when you’re composing your post so you can improve your reach.

AI Customer Service

• Customer service suite: Orlo’s dedicated AI product for grading and triaging inbound digital demand while protecting handlers from harm.
• Message classification: This auto-categorises inbound messages against your own taxonomy to prioritise and route messages. 
• AI image analysis: Created alongside Derbyshire Constabulary, our image filter automatically blocks explicit or abusive pictures from appearing on screen, protecting your staff from harmful content.

Social Listening and Media Monitoring AI features

• Emotion and Sentiment analysis: The system goes beyond basic positive or negative labels to pick up on specific emotions like anger or frustration. For example, a missed bin collection and a dangerous pothole both register as negative sentiment, but they require very different emotional responses. This helps to dig into why people are feeling certain emotions in order to build trust. 
• Theme Alerts: Standard monitoring tools only track words you explicitly tell them to search for. Our Theme Alerts spot unexpected topics bubbling up across social media and news sites automatically, and alerts your team early.
• Trust Score: By tracking sentiment and emotion over time, Orlo’s Trust Indicator calculates a simple Trust Score alongside SWOT analysis, making it easy to report real impact back to senior leaders.

• Policing: One of our metropolitan forces uses message classifications to identify emerging risks in real time across contact and comms, helping teams respond earlier, tackle misinformation and save staff time.
• Healthcare: Many of our NHS organisations use AI-generated alt text to make informative content more accessible to people with visual impairments and harder-to-reach communities.
• Local Government: One of our local councils used AI to quickly identify incorrect information in a press release, helping the team respond swiftly, correct the information and share updated comms.
• Housing and Transport: Comms and customer service teams use Orlo Insights to spot trends, local issues, stakeholder conversations and incidents in real time. It brings social and customer data together to quickly understand what’s happening and who’s impacted.

People spot generic, robotic text almost instantly, and when public messaging feels cold or artificial, it’s not surprising that residents stop paying attention. Software does not have local context, empathy, or “good taste”, as I put it! Those qualities rely entirely on experienced public sector communicators, and cannot be replicated. 

Our future roadmap focuses on cutting out tedious admin tasks while keeping your team in total control. We are currently working on:

• Action plans in Insights: From theme to owned action, tracked inside the report that found it. 
• Content moderation: Broader protection for your channels and the people who staff them.
• AI query builder: Ask a question in plain English in Orlo Insights and AI will build the social listening and media monitoring query for you, instead of you using boolean logic.

Using AI tools in the public sector is about giving your staff enough space to listen properly and make informed decisions.

By keeping a human in the loop that can be in  control, empathic, and make final decisions,, public sector teams can work faster while keeping community trust intact.