AI Agents & Communications – The audience is a robot now?

While generative AI (Artificial Intelligence) has created lots of headlines and buzz, it has yet to deliver on its promises. It costs a lot of money, is terrible for the environment and has not been adopted by the public in a meaningful way. It also plagiarises creative work and is being used to justify paying creatives less. It might even be making us stupid.

All that being said, it can be argued that AI agents, are what generative AI has been building towards. It is, potentially, a huge disruption. It offers tangible value to a wide range of people and companies. An iteration of existing technology that will change the way we have to think about product design, communications and marketing.

While we are in the early days of agentic AI and AI agents, this technology will be rapidly adopted by companies to drive internal innovation and improve efficiency. Public adoption rates will be much slower. But the second order implications of Agentic AI, on communications, have not been widely discussed.

So, let’s get that conversation going.

What is Agentic AI, AI Agents and Generative AI?

It is important to define Generative AI, Agentic AI and AI Agents before discussing the key differences between them.

Generative AI can create content such as text, images, videos or even software code to varying levels of quality in response to a human query (commonly referred to as a “prompt”). Most of the big generative AI platforms (such as ChatGPT and Copilot) work as a kind of probability engine, using large language models (LLMs), neural networks and machine learning to provide the most probable “best” answer to prompts from users.

While previous AI assistants were rules-based and could not act independently, even those with LLMs built into them, AI agents can perform tasks proactively. They can perform complex tasks with little or no human supervision.

AI agents are AI systems that can make decisions and act on behalf of the user. It is important to note here that agentic AI and AI agents are also different.

Agentic AI refers to the overall framework, including the technological infrastructure. AI agents refers to individual AI “agents” that can make decisions on behalf of a user.

It can be quite confusing, to say the least! But AI agents could be the biggest disruption to communications and marketing since the rise of Google and social media networks.

The shift

Traditional communications practices are targeted towards human interpretation and awareness. Future communications practices will still have to do that, but also help machines find content, read/ingest that content, take action, and respond to the user.

Communications professionals need to shift from a model that only speaks to humans, to a model that also integrates machine readable publishing practices. This is nothing new for anyone who has worked in SEO (Search Engine Optimisation), but it is a very different way of working for most people in comms.

It affects the whole communications ecosystem, from internal corporate communications to public relations through to the media. It is possible that every stage of the communications funnel will be disrupted by AI agents.

For example, PR outreach campaigns rely on journalists picking up pitches. Those journalists, if they are not already, will use AI agents to screen those pitches. The public audience for media outlets will use AI agents to help them consume content. Agents could dictate what people read, creating even more silos than our current landscape of social media platforms and algorithms.

Workplace AI agents will likely distribute corporate messaging, assign tasks, book vacation leave, and even interact with employee’s personal AI agent, on their device.

These scenarios, while hypothetical, are not far away once certain issues around AI agent protocols are solved.

Which means that communications professionals need to start planning how to communicate with agents as an audience.

Publishing for robots

We don’t know how AI and agents will evolve. We have an idea right now, we know protocols like A2A and MCP provide a framework for us to use and they could help agents work together to complete tasks. But once quantum computing becomes a reality, we really don’t know where things will go.

But what we can do in communications, is plan for the audience. A big part of that audience will, quite soon, be robots.

Communications professionals need to take three new factors into account.

We need to create content that agents can discover, has information that agents and humans need and enable the agent to take action. We also have to think about how we track that action.

So do we create one set of messaging for robots and one for people? Not quite.

For example, even now you have to make your blog posts accessible to search engines and social media algorithms, if you want anyone to read them. But you still have to write a good post so the reader enjoys it. Factors like emotional storytelling are still vital.

But comms pros will need a machine-readable format for these stories too.

In this respect, it is not vastly different to creating SEO-focused content. You don’t create content with SEO as the sole purpose of it, but you use SEO practices to make sure search engines can access it. A big part of technical SEO is making content accessible to bots; the same principles will apply to AI agent-focused content.

So while it is a bit of a shift in mindset, communications professionals still provide value as we are the messaging experts. We just have to talk to a new type of audience and that will require some more technical skills.

Why media still has a role to play

A recent MuckRack paper has highlighted that generative AI really leans on mainstream media mentions. both to research answers from a user prompt and to provide citations to users that give them more trust in the answers they get. Generative AI still gets things wrong all the time. Media links help to steer it away from giving daft responses.

Media citations in generative AI engines also have crisis comms implications. Say coporate brand X does something silly and releases a bad product which causes reputational damage. Social media might be full of angry customers and snarky stories about the brand. Generative AI will echo those stories again and again. But a good media strategy, getting that brand in the mainstream press, will start to correct the generative AI platforms. They will still mention the crisis in their responses but they will also cite stories from the mainstream press that correct the initial narrative.

This will become part of communications reporting. Comms professionals do brand media monitoring now, soon they will also report citations as a KPI (key performance indicator).

This holds true for AI agents too. Correcting a media narrative will steer agent response and action. In our example, brand X has had its crisis and the response has put them all over the mainstream press, correcting the previous narrative. If an AI agent is making a purchasing decision, it will take this into account and it will do it at the same time as looking at the crisis. So when it goes back to the user for permission to buy a product from brand X, it will talk about the crisis but assure the user it has passed.

This is the type of imaginary scenario that could become real one day. So communications professionals need to plan for them.

It also tells us that communications are a key stakeholder, even in this time of disruption. If anything, communications professionals could be busier than ever. One of the core elements of corporate comms is getting a client to tell their story and get that story out there into the media.

Generative AI engines rely on citations from high authority media outlets to not only generate responses and steer away from errors. They use also use them to generate trustworthy responses. While some segments of the public have a deep distrust of the media, these citations still provide an element of digital proof. According to the MuckRack paper “more than 95% of links cited by AI are from non-paid media”. Media links assure the user that the AI is not making things up, it’s almost the model showing its workings. We can speculate that AI agents will have to do something similar for users on some decisions.

So in the coming years, some of the audience may be robots. But the importance of great media management, earned media and getting the story out there for clients remains.

Key takeaways:

• Media strategy is a crucial element of the AI-optimised future of comms
• Getting stories placed in high-authority media outlets is going to be even more important
• Start tracking brand citations in generative AI engines as a key success KPI to show to clients
• Correcting AI narratives will be part of crisis comms. Start fitting this into your plans