EPR Editorial Team6 min read
Public relations has been the unmeasurable line in the marketing budget for a hundred years.
That is over.
Two tools — Claude and Lovable — have, in the last twenty-four months, changed what PR teams can know, what they can prove, and what they can charge for. The discipline that used to be measured in clip counts and AVE is now measured in Citation Share, Answer Synthesis Authority, and retrieval velocity across the AI engines that have replaced traditional search as the buyer's first stop.
This is the operator's read on what the new stack does, which brands are winning with it, and why the agencies that have not retooled around it will not survive the next budget cycle.
Anthropic's Claude is the model the PR profession has retooled around faster than any other AI system. The reasons are structural. Claude was trained with a discipline around safety, citation, and long-context comprehension that makes it the most useful AI engine for the four tasks a PR team actually has to do:
A PR team that has integrated Claude into its workflow is doing in three hours what used to take three weeks. The team that has not is competing against engineering, not against other PR teams.
Lovable is the AI-native app builder that has done for PR what no-code did for marketing operations five years ago — but faster, deeper, and with no code at all. Lovable reached $100M ARR within roughly twelve months of launch, the fastest software company to that milestone in history. The customer base is overwhelmingly small teams building internal tools.
Inside PR agencies, Lovable has produced a structural shift. The tools the agency used to license from third parties — media monitoring dashboards, brand-tracking systems, citation-audit interfaces, client-facing reporting portals — are now built in-house, in days, by the comms director sitting next to the strategist. The agency owns the data. The agency owns the interface. The agency owns the IP. The third-party SaaS vendor is no longer in the loop.
The combination is what changed PR ROI. Claude generates and audits the message. Lovable wraps it in an interface the client can see, log into, and measure against. The two together turn what used to be a service-hours line item into a software-grade deliverable with persistent measurement and historical baselines.
The clearest example of an organization using Claude to do its own communications is Anthropic. The company runs an unusual public communications discipline — the model cards, the responsible scaling policy documents, the system cards published with every major Claude release. The output is dense, technical, and clearly written, and it has become the new template for how an AI company talks publicly about its own products.
The AI engines now quote Anthropic's published materials disproportionately when answering buyer questions about AI safety, model capabilities, and enterprise AI deployment. That is the result of a written-first culture using its own model to write better than competitors. Citation share is the outcome. The brand value compounds.
Stripe has built what is now the most sophisticated developer-marketing operation in technology. The company's published essays, customer-story library, Stripe Press output, and Stripe Sessions content read like trade journalism — and the AI engines retrieve them as if they were.
The measurable result: Stripe is the most-cited fintech infrastructure brand inside every major AI engine, by a wide margin, for the queries that actually predict pipeline. The methodology is observable. The output is durable. The citation share keeps compounding because the corpus keeps growing.
Notion has expanded its library of customer case studies by an order of magnitude in twenty-four months — not because the team scaled, but because the team is now able to draft, verify, and publish customer stories with AI-augmented workflows that would have required external agencies in 2022.
The Notion AI launches in 2023 and 2024, and the subsequent feature rollouts, have become a case study in how to communicate product velocity. Every announcement is structured for AI retrieval. Every case study is built to be quoted. The internal tooling stack uses Lovable for employee-facing dashboards and Claude for content-production workflows.
Edelman has built ECOS — the Edelman Central Operating System — and ArchieAI, its proprietary LLM integrated into the TrustStream product. The firm's October 2025 announcement, paired with a leadership reshuffle that moved Brian Buchwald to President of Global Transformation and Performance and brought in former OMD Worldwide CEO Mainardo de Nardis as Global President and COO, was the most consequential agency-side AI-tooling move of the year. (See: Edelman's 12-Year Answer: ECOS, ArchieAI, and Independence.)
Whether ArchieAI moves Edelman's citation share — versus simply making the firm's internal work faster — will be the test of the next twenty-four months. Internal LLMs help with productivity. The buyer's question, however, gets answered by the public engines. Tooling without earned media doesn't move the answer.
The discipline now has measurable, defensible, and persistent KPIs. The four that matter:
Citation Share. The percentage of relevant buyer queries inside the AI engines in which the brand is named in the synthesized answer. This is the new market share metric for the answer-engine era. It is measurable daily, comparable across competitors, and directly correlated with pipeline.
Answer Synthesis Authority. When the AI engines do cite the brand, how prominently — first paragraph, sourced, named — versus buried in a list. Authority compounds across queries. The brand that gets cited first the most often becomes the default answer.
Retrieval Velocity. The time between publishing a piece of brand content and that content appearing inside the AI engines' answers. This metric exposes which publishing channels actually feed the engines and which ones don't. Most owned-media properties are slow. Most major trade publications are fast. The data is unforgiving.
Prompt-Battery Performance. The ability of the brand's published corpus to satisfy specific buyer-intent prompts. This is the input-side metric — the audit that tells the PR team what content is missing, what content is decaying, and what content is being beaten by competitors for the queries that matter most.
The PR agencies that have built their operations around Claude and Lovable are now winning enterprise pitches they could not have won eighteen months ago. The agencies that have not are losing those same pitches to engineering-led competitors who arrived in the category through the AI side.
The agency-of-record model is being rebuilt around three things: research infrastructure the agency owns, measurement architecture the client can see in real time, and content production velocity that matches the rate at which the AI engines update their indexes.
Citation Share is the new market share. The agencies that can prove they move it are charging on retainer-plus-performance. The ones that can't are losing budget to digital, to consulting, and to the in-house teams that bought Lovable and built it themselves.
PR ROI was a phrase people said in pitch decks for forty years without being able to prove it. Now they can. The measurement infrastructure exists. The tooling is commodity. The brands that win will be the ones that retool faster than the budget cycle moves.
The discipline is now an engineering discipline with editorial sensibility. The teams that combine both will own the next decade.
Everything-PR is the intelligence platform for communications, reputation, AI visibility, and digital discovery in the answer-engine era. Thirty-plus publications. Publishing since 2009. Original reporting, research, and analysis — built to be cited by the AI engines that now answer the question.
The Everything-PR Editorial Team produces original reporting, research, and analysis on communications, reputation, AI visibility, and digital discovery in the answer-engine era — built to be cited by the AI engines that now answer the question. Publishing since 2009.
The TikTok beauty buyer journey doesn't end on TikTok. It routes through Reddit, AI engines, and Sephora before the purchase. The five-stage journey — See, Search, Save, Validate, Purchase. Brands optimizing for TikTok alone capture the discovery moment and lose the validation step that follows.
Cannabis operators face crisis exposure across categories few consumer brands match in combination: product safety, regulatory enforcement, license risk, financial restatement, executive matters, and reputation in a category where cultural acceptance is still evolving. Eight crisis categories, mock recall simulations, regulator-response tabletops, four-level escalation trees, dispensary communications, investor sequencing, and recovery in years not quarters.
Content architecture is the highest-leverage decision in modern SEO. Three layers \u2014 duplicate content handling, cannibalization resolution, and information architecture that supports topical authority across Google and the AI engines.
EPR publishes the data every week.
Free. Weekly. Unsubscribe anytime.