
Sedestral links its SEO agents to Claude and ChatGPT via MCP
Sedestral has connected its SEO agents to Claude and ChatGPT through the MCP protocol, according to Abondance. The report does not detail which tasks are automated, what the integration costs or how access is controlled.
Abondance reported on 3 November 2026 that Sedestral has connected its SEO agents to Claude and ChatGPT through the MCP protocol. MCP, short for Model Context Protocol, is the open standard that lets an AI assistant call external tools and data sources rather than working only from what is pasted into its chat window. The report frames the move as a shift in how SEO work is driven day to day; it does not say which tasks the agents now cover, how the connection was built, what it costs, or how client data is protected.
What the announcement actually changes
The report states that many SEO practitioners already spend a large part of their day inside Claude or ChatGPT: reading a Search Console export, drafting a brief, checking a title tag, assembling a client report. Once an assistant can reach a platform's own tools through MCP, that work no longer has to start with a manual export and end with a manual copy-paste. The practical gain is not a smarter model; it is fewer handoffs between a dataset and the place where decisions about it get made. What remains unknown is the depth of access Sedestral grants, whether the agents can write back to the platform or only read, and how a client's search data is handled once it passes through a third-party model.
Why agencies should care
For teams selling or buying SEO, the interesting question is where the human review sits. An assistant that can query Search Console directly will produce answers faster, but the same assistant will also produce confident mistakes faster if nobody checks the underlying query. The teams that get value from this pattern tend to define, before wiring anything up, which outputs require a second pair of eyes and which can pass straight through. The report does not specify how Sedestral handles that, so treat the question as open for anyone assessing similar setups.
There is also a commercial dimension. If analysis and reporting become cheaper to produce, the billable unit shifts from hours spent compiling to judgement applied on top. Buyers should expect to see that reflected in how SEO retainers are structured, though nothing in the announcement says Sedestral has changed its own pricing.
What to check this week
- List the three recurring SEO tasks that currently require a manual export; these are the first candidates for a tool-connected assistant.
- Ask any vendor claiming MCP or similar integrations whether access is read-only or read-write, and where client data is stored.
- Pick one report your team produces weekly and time it; that baseline is what an integration should be measured against.
None of this makes the underlying SEO work disappear. It moves the bottleneck from producing an analysis to deciding whether the analysis is right, and that decision still belongs to a person who knows the site.
Also worth reading
-
llms.txt: what the file does and does not do for AI visibility
WebRankInfo examines llms.txt, a proposed file meant to help site publishers interact with AI assistants such as ChatGPT and Gemini. The report argues the file is widely misunderstood and currently of no practical use.
-
Blocking ChatGPT and OpenAI crawlers: what it means for your traffic
WebRankInfo has published guidance on blocking OpenAI and ChatGPT from using a site's content to train models and generate answers. For publishers, the decision trades potential future visibility in AI answers against the risk of losing attribution and traffic.
-
New York Times sues Brave over articles used for its AI features
The New York Times has filed a complaint against Brave, alleging its articles were used and exploited without authorisation, including to feed the browser's AI features. The publisher joins a broader front of media organisations challenging how AI companies use their content.


