AI Citations: PR Agencies Rush Into GEO Partnerships

A new middleman between PR and the chatbot

Public relations agencies are lining up behind a new acronym. In 2026, a series of partnerships has paired traditional communications shops with firms that specialize in tracking and improving how brands show up in AI-generated answers. The pitch is consistent. Generative engines like ChatGPT, Claude, Gemini and Perplexity increasingly sit between consumers and the brands they research, and the material those engines cite is overwhelmingly earned media.

In May, Baltimore-based Abel Communications, an integrated agency focused on earned media, thought leadership and content strategy, joined Brandi AI's global Agency Partnership Program 3. Under the arrangement, Abel will use Brandi's tooling to show clients how they are surfaced, cited and discussed across AI platforms, then turn that insight into credibility-building work 3. In October, AI visibility consultancy SearchTides and boutique agency Needle announced a preferred partnership aimed at helping clients earn placement in generative search answers 4. The industry calls that goal answer engine optimization, or AEO.

The two deals follow the same template. A firm with media relations skills teams up with a firm that measures AI outputs. Each side supplies what the other lacks.

The data driving the deals

Much of the case rests on research from Muck Rack, whose Generative Pulse team analyzed more than 25 million links pulled from ChatGPT, Claude and Gemini responses across 17 industries 1. The May 2026 edition found that earned media accounts for 84% of AI citations 1. Paid and advertorial content made up just 0.3%, and journalism alone represented 27% of cited sources 1. An earlier Muck Rack analysis, discussed in a March webinar, put the earned-media share at 82% 2. Across reports, the figure is stable or edging up, not eroding.

The report also stresses that each platform behaves as its own information environment. ChatGPT, Claude and Gemini differ markedly in which sources they lean on 1. For communications teams, that complicates measurement. Being visible in one assistant does not guarantee visibility in another.

Muck Rack's product and data leads have framed this as continuity, not disruption. Senior director of product Brian Maehl argued that the work of PR itself isn't changing. In his view, GEO is a new "wrapper" around familiar work, and that work now carries more influence 2. The practical advice is equally familiar. Teams should measure share of voice inside LLM responses the way they once measured it in press coverage, while recognizing that results shift with how a question is phrased 2.

The ownership gap

If the data is clear, the organizational response is not. Muck Rack's 2026 State of PR report found that nearly a third of PR professionals say no one at their organization owns generative engine optimization 4. SearchTides and Needle explicitly position their partnership as a fix for that gap 4. Abel and Brandi frame theirs around building trust, recognition and visibility in AI answers 3.

This looks like a land grab in its early phase. SEO consultancies, analytics vendors and PR agencies all have a plausible claim to GEO. Partnerships let each hedge without building the full stack in-house. Agencies get dashboards that make AI visibility reportable to clients. Visibility firms get access to the earned-media machinery that, by Muck Rack's numbers, actually moves citations.

Reading the signals with care

Some caution is warranted. The most widely cited statistics come from Muck Rack, a company that sells media relations and PR analytics software. It has a commercial interest in showing that earned media matters to AI systems. That doesn't invalidate a 25-million-link sample, but the research has not been independently replicated. The partnership announcements are promotional by nature. They describe intended capabilities, not demonstrated outcomes 34.

The methodology question also matters. Muck Rack itself notes that prompt wording shapes results 2. Platforms differ substantially in sourcing behavior 1. Any vendor promising "AI visibility" is measuring a moving, model-specific target that can change with each model update.

What it means

The strongest takeaway is less about tools than about incentives. If paid placements barely register in AI citations 1, brands can't simply buy their way into chatbot answers. They have to be written about by credible third parties, especially journalists. That rewards the slow work of pitching relevant stories and building media relationships.

The partnerships multiplying across the industry are a bet on that dynamic. They suggest PR's value proposition is being repackaged for an AI audience rather than replaced. Whether the new measurement layer delivers results beyond what good media relations already provided is still unproven.