Claude (Anthropic)

Claude is a family of large language models and an AI assistant built by Anthropic. People use it through the claude.ai apps, through tools such as Claude Code, and through the Claude API, where developers call the models from their own scripts. For SEOs it matters twice: as a working tool for analysis and automation, and as an AI search surface that can find, read and cite your pages.

How it works

Claude models are offered in tiers that trade capability against speed and cost. As of October 2026, Anthropic’s documentation lists Fable, Opus, Sonnet and Haiku models, with Haiku positioned for high-volume tasks such as classification, extraction and routing. Current models accept text and images, support tool use and have long context windows. Model names and specs change often, so check Anthropic’s models overview before choosing one.

Claude can search the web and cite sources. Anthropic documents three crawlers that site owners can control separately in robots.txt: ClaudeBot (collects content that may be used for training), Claude-User (fetches pages when a user asks) and Claude-SearchBot (improves search result quality). Anthropic also created the Model Context Protocol, which lets Claude and other assistants connect to tools such as SEO data APIs.

How it’s used in SEO and marketing

• Classifying keywords, pages or SERP features at scale through the API.
• Drafting anchor text, summaries or briefs inside scripted pipelines, with guardrails and human review.
• Running agents that sit on top of deterministic scripts and explain what changed.
• Tracking brand visibility: testing a fixed prompt set in Claude alongside ChatGPT, Perplexity and Gemini.
• Measuring AI referrals: visits from claude.ai can be captured in a GA4 “AI assistants” channel.

Example

MLforSEO’s internal linking script uses the Anthropic API (Claude) by default to suggest anchor text, and swapping in another provider means changing one function. In the SERP analysis guide, running every AI step live (classification, question clustering and the agent) cost about 17 cents of Claude credit in total.

Limitations and things to watch

• Like any LLM, Claude can produce hallucinations. Ask for evidence and spot-check outputs.
• Newer Claude models no longer accept sampling settings such as temperature; Anthropic recommends steering behaviour with the prompt.
• Pin model IDs in scripts so results stay comparable over time, and expect older models to be retired.
• Blocking ClaudeBot does not block Claude-User or Claude-SearchBot. Decide on each separately.

Learn it on MLforSEO

• 10 Ways to Automate Internal Linking with AI
• 10 Ways to Automate SERP Analysis with AI
• Your robots.txt might be the silent killer of your AI traffic
• AI Search Optimisation & Agentic SEO course, lesson on language model visibility fundamentals.