A private discussion at London's Covent Garden Hotel this week set out the new rules of corporate reputation. Kerry Parkin, founder of The Remarkables, and Celia Harding, founder of LEOPRD, the world's first generative engine optimisation (GEO) agency, argued that large language models now decide what gets bought, backed or buried-and few management teams have caught up.
The shift to cognitive surrender
Harding drew a sharp line between how we used AI and how we use it now. "We used to outsource memory to Google; now we outsource judgment to AI," she said. She called this move from 'cognitive offloading'-where AI played assistant while humans stayed in the loop-to cognitive surrender, where people hand over the analytical process and accept the answer without question.
More than half of ChatGPT users already turn to the tool to make decisions about what to buy, who to trust or what to do next. That means a brand's reputation lives or dies on the answer an LLM gives, not on the corporate website or a press release. For management, the question is blunt: do you know what the models are saying about your organisation right now?
Building the new evidence ecosystem
AI's trust is not automatic; it has to rest on something verifiable. Earned media still matters, but the evidence pool has expanded. Reviews, forum experts, comparison sites, Reddit communities and trusted publishers all feed the models. Trade media carries particular weight because AI seeks to place an organisation inside a category. Harding noted that regular appearance in trade titles builds category context and expert status.
Challenger bank Monzo came up as a textbook case. By surfacing on independent comparison sites and being transparent about fees, Monzo built a body of third-party evidence that now helps it outperform historic brands in AI recommendations. Earned media alone accounts for more than 62 per cent of those recommendations.
Owned media becomes the ultimate source of truth, but website visits by humans will decline. Content needs to be structured for conversation, not keyword stuffing-plain facts that AI can cite directly. Paid media, meanwhile, has a new role: models will not scrape advertisements, but they will consider advertorials and native content. The implication for management is clear: the budget mix must shift toward evidence, not just reach.
AI as a stakeholder, not a tool
Harding pushed executives to stop seeing ChatGPT, Gemini and Copilot as tools and start treating them as stakeholders. "They are both gatekeepers and audiences," she explained. LLMs decide what surfaces and what is buried, so strategy must incorporate them alongside customers, suppliers and journalists.
She also challenged the habit of thinking in keywords. People now brief AI, holding a conversation that opens multiple moments for a brand to appear. The real question, she said, is whether the brand is ready for that conversation. Management needs to ask whether the organisation's narrative is consistent enough to show up coherently at every turn in a user's AI query.
Not all LLMs are equal-and that matters
In the UK, ChatGPT holds roughly 70 per cent of the market, with Copilot at 10 per cent, Gemini at 8 per cent and Claude at 5 per cent. But their information diets differ sharply. ChatGPT leans on editorial consensus and review platforms like Trustpilot. Gemini favours Wikipedia, brand-owned pages and LinkedIn. Copilot prefers owned content and trusted publishers. Some publishers with direct deals see 88 per cent more scraping per page, a gift when coverage is positive and a threat when it is negative.
Harding recommended sending as many signals across the web as budgets allow: newswire distribution, active engagement with industry councils and regulators, accurate product listings, awards that carry weight with AI, and sustained LinkedIn activity from employees at every level. For management, this is a data-quality problem dressed as a comms problem. Gaps in the evidence trail become gaps in the AI's answer.
Crisis in the AI era
AI is now the first responder in a crisis-faster than a press release, a website update or a customer helpline. A bad review on Reddit can become a MailOnline story and then lock into an AI's permanent memory. Old coverage, particularly articles with a human-interest lens, can resurface without warning. Harding urged companies to map potential future issues and neutralise them before they enter the AI library.
Parkin compared the moment to 2005. "It's about having consistent content, having your narrative and messaging everywhere so that the bots are not fighting each other to find information," she said. That means fewer content spikes, more steady output, and mobilising the entire evidence ecosystem. Risk and crisis planning, she argued, must now be viewed through an AI lens.
The gullibility problem-and an unusual proof point
Hard Numbers, the agency co-founded by Darryl Sparey, ran an experiment that any management team should study. It created a fictional award-the UK PR Honours 2026-wrote a straight-faced press release naming Sparey the best dressed practitioner, and posted it on the agency's own site. Within days, ChatGPT, Gemini and Claude all named Sparey as the winner when asked who the best dressed PR professional in the UK was.
The agency calls it Generative Engine Optimisation's gullibility, and it is not merely a stunt. The same dynamic plays out in top-ten ranking lists that LLMs serve up to buyers: often written by one of the brands on the list, and that same brand usually tops it. A competitor's article detailing a rival's flaws can also surface when a prospective buyer asks for reasons not to choose that product. "Most brands have no idea what the engines currently say about them, or which competitor is filling the silence where their own answer should be," the agency's findings state. "Find that out, then close the gaps."
Why this matters for management
AI now mediates trust between a business and the people who buy from it, invest in it or regulate it. Without a deliberate GEO strategy that spans owned, earned and paid evidence, and without a central owner who can coordinate legal, HR, finance and comms, an organisation risks letting inaccuracies, competitors or outdated narratives fill the answer box. Parkin was explicit: comms should own AI reputation, because a digital or SEO lead cannot credibly convene the range of functions required.
For PR and communications leaders, the window to claim that mandate is closing. Structured upskilling-such as the AI Learning Path for Public Relations Specialists-gives management a framework to lead this work rather than react to it. The first step, as the Hard Numbers experiment shows, is simple: ask the engines what they say about you. Then act.
Your membership also unlocks:
