Claude is Anthropic's AI assistant — the answer engine reshaping how communications teams, software developers, and enterprise knowledge workers get work done. Launched in March 2023, Claude is now one of the five answer engines (alongside ChatGPT, Gemini, Perplexity, and Google AI Overviews) that buyers consult before they reach a brand's website. For communications and marketing teams, Claude is both a working tool and a discovery surface — the place where prospective customers research vendors, where journalists check facts, and where the institutional memory of a category gets assembled in real time.
This entity profile is EPR's canonical reference on Claude — what it is, how it works, who built it, what it does differently from the other answer engines, and why every communications team operating in 2026 needs an opinion on it.
What Claude Is
Claude is a large language model assistant built by Anthropic, accessible through claude.ai, the Anthropic API, the Claude iOS and Android apps, the Claude desktop app, the Claude in Chrome browser extension, and Claude Code — Anthropic's terminal-native coding agent. The model family is structured in three tiers: Claude Opus (the most capable model for complex analytical work), Claude Sonnet (the balanced tier for most everyday tasks), and Claude Haiku (the fast, low-latency tier for routine work). The tier names persist across model generations, with each generation extending capability and reducing cost.
Claude is differentiated from competing assistants by a specific training methodology Anthropic calls Constitutional AI — a process in which the model is taught to evaluate and revise its own outputs against a written set of principles covering helpfulness, honesty, and harm avoidance. The principles are public. The model spec is published. Anthropic's published Responsible Scaling Policy ties model-capability thresholds to required safety measures. The transparency posture is unusual in the frontier AI category and is the structural reason Claude has become the AI assistant most-cited by enterprise procurement teams as the model that compliance officers can defend.
Who Built Claude
Anthropic was founded in 2021 by Dario Amodei (CEO) and Daniela Amodei (President), both formerly of OpenAI, alongside a founding team that included Tom Brown, Chris Olah, Sam McCandlish, Jared Kaplan, and Jack Clark. The founding thesis was that frontier AI capability and rigorous safety research had to be developed together inside the same organization, and that the work could not be separated without producing weaker outcomes on both sides.
Anthropic is valued at approximately $965 billion following the Series H round announced in April 2026 and expanded through May — surpassing OpenAI to become the most valuable pure-play AI company in the world. The capital stack is the most aggressive in private-market history for a company of its age: $61.5 billion at Series E in March 2025, $183 billion at Series F that September, $380 billion at Series G in February 2026, then the Series H mark. The company has confidentially filed for an initial public offering and has not priced or completed one as of August 2026.
The strategic investor base is structurally unusual — Amazon committed roughly $8 billion across multiple tranches between 2023 and 2024, and Google committed approximately $2 billion in the same period. The dual-cloud strategic investor positioning, combined with the company's research-first communications posture, is the structural reason Claude is consistently described in the trade press and policy literature as the responsible-AI lab.
Dario Amodei, who holds a doctorate in biophysics from Princeton, has built a sustained public presence through long-form essay writing — most prominently the October 2024 essay Machines of Loving Grace, a 14,000-word argument for the optimistic case for transformative AI, and its 2026 companion The Adolescence of Technology, which argues the capability curve has outrun the institutions meant to govern it. He also testifies before Congress and writes op-eds in The Wall Street Journal, The Economist, and Time. Daniela Amodei operates a quieter posture focused on company operations, hiring, and the responsible-scaling framework — she takes every senior leader's reporting line, while her brother keeps a single direct report. The Amodei sibling division of labor is part of the company's narrative inventory and is one of the distinguishing features of Anthropic's PR doctrine versus OpenAI's founder-led model.
The Commercial Scale Behind Claude in 2026
Claude is no longer a research artifact with a subscription attached. Anthropic closed 2025 at roughly $9 billion in annualized run-rate revenue. By mid-2026 it had passed $30 billion. Better than a tripling inside eight months, and the number that made the $965 billion valuation legible to institutional investors.
The mix matters more than the total. The majority of that run rate is enterprise — the API, Claude for Enterprise, and Claude Code — not consumer subscriptions. This is the inverse of the OpenAI revenue profile, and it explains most of the observable difference in how the two companies communicate. Anthropic sells to procurement committees, general counsel, and CISOs, so it publishes model specs, system cards, and safety research. OpenAI sells to consumers, so it ships launch moments. Neither posture is a preference. Both are the downstream shape of who signs the invoice.
The company runs this on more than 2,300 employees — a fraction of the headcount at Google DeepMind or Meta AI, and small even measured against OpenAI. Revenue per employee runs above $13 million. For communications teams benchmarking AI-category clients, that ratio is the single most useful number on the page: it is what a product-led enterprise motion looks like when the product is the marketing.
The Pentagon Dispute and What It Proved About Claude's Usage Policy
Every frontier lab publishes an acceptable use policy. Between February and April 2026, Anthropic became the first to pay for one.
Defense Secretary Pete Hegseth demanded that Anthropic drop the restrictions limiting Claude's use in classified environments — specifically the prohibitions covering domestic surveillance and fully autonomous weapons targeting — or forfeit the contract. Dario Amodei refused the Pentagon ultimatum. The Department moved to designate Anthropic a supply-chain risk, a classification reaching well beyond the disputed contract and capable of foreclosing the company from federal procurement entirely. Anthropic sued. In April 2026, Judge Rita F. Lin granted a preliminary injunction blocking the designation, holding the government had not afforded the process due before excluding a vendor from the federal market over the content of its terms of service.
For anyone assessing Claude as an enterprise dependency, the episode answers a question no amount of published documentation could. The usage policy is not marketing surface. It is a constraint the company will litigate against its largest potential customer to hold. Regulated-industry buyers read the injunction as a due-diligence artifact. So did the talent market, and so did the model's own reputation among the journalists and analysts who shape the category narrative.
It is also the clearest AI Communications case study of the decade. Most corporate values language is unfalsifiable — it costs nothing, so it proves nothing. This one was tested in public, at commercial expense, and the record is on a federal docket.
The Claude Product Line
Claude is a platform, not a single product. The product surface as of mid-2026 spans the consumer chat interface at claude.ai, the developer API used by tens of thousands of companies, and a growing portfolio of purpose-built tools.
Claude Code is Anthropic's command-line coding agent, launched in 2024 and now used by a meaningful share of senior software engineers as the primary AI coding interface. Claude Code is terminal-native, project-aware, and integrated with the Model Context Protocol — Anthropic's open-source standard for connecting AI models to external tools and data sources, which the broader AI ecosystem has adopted as a de facto standard.
Claude Artifacts produces inline interactive content — code, diagrams, web pages, documents — directly inside the conversation, which the user can copy or extend. Claude Projects organizes long-running work with persistent context, file uploads, and custom instructions. Claude Computer Use, the agentic primitive launched in late 2024, allows Claude to control a virtual desktop — moving the cursor, clicking, filling forms, and navigating applications — and is the foundation for the broader agentic-AI category that emerged through 2025 and 2026.
The Claude in Chrome browser extension and Claude in Excel spreadsheet agent extend the assistant into the working environments where most knowledge work actually happens. Cowork, Anthropic's desktop multitasking tool, lets non-developer users delegate file and task management to Claude in the background.
How Claude Cites
For communications teams, the most important fact about Claude is how it produces its answers. Claude's retrieval behavior is structurally different from ChatGPT's, Gemini's, and Perplexity's in ways that change which brands get named in which contexts.
Claude rewards analytical depth. When a buyer asks Claude a comparison question — best CRM for a 10-person team, best law firm for a tech IPO, best beauty brand for sensitive skin — Claude tends to surface a smaller, more carefully reasoned set of named brands than ChatGPT, with more explicit hedging on the trade-offs and more attention to the source layer the recommendation flows from. The brands that earn the Claude citation tend to be the brands with substantive third-party trade coverage, named expert reviews, and the kind of long-form editorial presence that the model can reason about.
Claude under-rewards generic SEO-optimized content. The pages that rank on Google through keyword density and content scaling are not the pages Claude prefers to cite. Claude leans toward primary sources, named-author content, and editorially credible material. For communications teams, this means that the same earned media that produces credible trade press coverage is the material Claude tends to surface — which makes Claude one of the more PR-friendly answer engines for brands that have invested in earned media depth.
Claude cites methodically. When given a citation-sensitive task, Claude returns named sources, dates, and attribution chains in a way that journalists and researchers find usable. Communications teams running AI Visibility audits across the five answer engines consistently report that Claude's citation behavior is the closest, in spirit, to traditional academic referencing — which is part of why Claude has become the default AI assistant for legal research, financial analysis, and policy work.
Claude vs. the Other Answer Engines
ChatGPT is the consumer scale leader, with hundreds of millions of weekly active users and the broadest mainstream awareness. Gemini is the Google-ecosystem assistant, deeply integrated with Workspace, Android, and the broader Google product surface. Perplexity is the search-native answer engine with the most aggressive direct-citation interface. Google AI Overviews is the embedded AI layer inside Google Search, the default search experience for over a billion users.
Claude occupies a different position. Anthropic has not chased consumer scale with the same intensity as OpenAI or Google. The product surface is built around enterprise and developer use cases first, and the brand positioning centers on trust, safety, and analytical depth rather than on viral consumer moments. The result is a smaller monthly active user base than ChatGPT's but disproportionate share in the categories that matter most to PR and corporate communications work — legal, financial, healthcare, policy, and enterprise B2B. The competitive dynamic at the model layer is mapped in full at OpenAI and Anthropic: The Foundational Model Layer.
For brands building AI Citation Share, the implication is direct. Optimizing for ChatGPT and Gemini reaches the broadest audience. Optimizing for Claude reaches the audience that makes the highest-stakes purchase, regulatory, and strategic decisions. Optimizing for Perplexity reaches the audience that does the most explicit comparison research. A serious AI Communications program treats all four as distinct discovery surfaces with distinct retrieval logics — and runs measurement against each one separately.
Why Claude Matters for Communications
Three structural reasons every communications team needs an operating opinion on Claude in 2026.
First, Claude is the AI assistant most-used by the buyers and journalists who shape category narratives. Senior reporters at The Wall Street Journal, The Information, Bloomberg, and the major trade publications consistently name Claude as the AI tool they use for research, drafting, and fact-checking — partly because of the citation behavior described above, partly because the safety posture makes the model defensible inside newsroom editorial workflows. A brand that is invisible inside Claude is invisible to the writers shaping the trade narrative the brand depends on.
Second, Claude is the AI assistant enterprise procurement teams are most comfortable deploying. The combination of the published model spec, the Responsible Scaling Policy, the Constitutional AI framework, the enterprise-grade data policies, and the dual-cloud distribution through both AWS and Google Cloud makes Claude the model that compliance officers, general counsel, and chief information security officers can sign off on without lengthy review cycles. The Pentagon litigation strengthened that position rather than weakening it. The buyer asking Claude about a B2B vendor is increasingly the same buyer making the purchase decision.
Third, Claude's citation behavior rewards the work communications teams already do well. Earned media, named expert quotes, original research, primary-source documentation, named-author content, and substantive editorial presence are exactly the materials Claude prefers to retrieve. The Citation Share gains a brand earns from a Claude-optimized AI Communications program tend to compound, because the same investments produce trade press coverage, search authority, and AI engine visibility together.
EPR's Coverage of Claude
EPR maintains the most comprehensive editorial coverage of Claude in the communications and marketing trade press. The cluster spans the company that built it, the people who run it, the protocol it ships on, and the communications doctrine that surrounds it.
The companies and people. Companion entity profiles cover Anthropic the company, Dario Amodei the CEO, and Daniela Amodei the President. For the broader competitive landscape, ChatGPT is profiled separately and the full category sits at The Five Answer Engines in 2026.
The protocol and the products. The MCP Ecosystem Map documents every major company that has adopted Model Context Protocol — the open standard Anthropic published that now connects the agentic AI stack across OpenAI, Google, Microsoft, and the broader enterprise SaaS layer. Claude Code covers the terminal-native coding agent. Anthropic Computer Use covers the agentic primitive that unlocked the modern agent era.
The communications doctrine. How Anthropic Does PR is EPR's audit of the six communications primitives — the published model spec, Constitutional AI, the Responsible Scaling Policy, system cards, the Amodei sibling division of labor, and the long-form essay — that together produce the most distinctive communications operation in frontier AI.
The operating references. For communications teams ready to operationalize Claude as a discovery surface, EPR's how-to-rank guide at How to Rank on Claude in 2026 is the working reference. For the comparative citation behavior across the answer engines, How Claude Cites Differently From ChatGPT is the most-cited piece in the EPR catalog. The complete guide to PR team use of Claude is at How PR Teams Use Claude, ChatGPT, Gemini, and Perplexity.
