Rhizome AI

Ask FDA, EMA and other health-authority questions from an AI agent and get cited answers.

Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.

Paste it into any MCP client. Setup docs

What the Rhizome AI connector does

The Rhizome AI MCP server connects Rhizome's regulatory research agent to Claude and other MCP clients. Ask a question about FDA, EMA or other health-authority positions, and the agent searches regulatory databases and returns an answer cited to source pages. Ten tools start and read those research conversations or search documents directly. Sign-in is OAuth.

Verified connector

Listed by Anthropic as a partner connector in its Connectors Directory.

Connection checked by Agentman on .

Anthropic states this reflects the level of review a connector received, not a security audit.

Rhizome AI tools (10)

  • rhizome_conversation_message
  • rhizome_conversation_wait
  • rhizome_conversation_read
  • rhizome_conversation_list
  • get_rhizome_conversation
  • rhizome_document_read
  • rhizome_search
  • rhizome_dataset_list
  • rhizome_conversation_display
  • rhizome_conversation_stop

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

  • Answers need checking. Rhizome states AI is still a probabilistic technology and that it does not take liability. Its Claude guide says to verify before you co-sign. Rhizome is also not for medical advice to patients, by its own description.
  • Slow by design. A research run usually takes one to ten minutes, per Rhizome, and there is no quick mode.
  • Free tier is small. Ten messages a month on the free plan, per Rhizome's FAQ. Rhizome's REST docs say a message over the free limit is refused with HTTP 402.
  • Many datasets are Enterprise-only. Rhizome lists Japan, China, South Korea, India, Brazil, Argentina and other national authorities, plus FDA MAUDE and FAERS, as Enterprise tier.
  • Concurrency is capped. Rhizome's REST docs allow three running conversations per account, counting those started on the web. Rhizome does not say whether MCP-started runs share that cap, but they appear in the same account.
  • No per-tool MCP documentation. Rhizome documents its REST API, not the MCP tools. Tool effects on this page are mapped by name to REST operations, and we could not read schemas or annotations because the handshake was refused without credentials.
  • Consent is identity-only. The two advertised scopes cannot limit what the connector does, so access is all or nothing at the account level.
  • Private systems are a separate project. Rhizome's Claude guide says connecting internal systems such as Veeva, RIM or SharePoint is a separate integration conversation.

Frequently asked questions

Does Rhizome train AI models on my questions or documents?

No, according to Rhizome's FAQ. It says questions and documents are used only to serve your request. Rhizome adds that its AI vendors may retain inference data for up to 30 days for abuse monitoring under standard API terms, then delete it, and that this data is never used to train models. Your AI client's own retention terms apply separately to the conversation.

Can I rely on Rhizome for a regulatory submission?

Not without checking it, and Rhizome says so itself. Its FAQ positions the product as doing the research underneath a decision, with every claim linked to a primary-source passage your team can verify in a click. It also states that AI is still a probabilistic technology and that Rhizome does not take liability. Treat answers as cited research, not as sign-off.

Is there a free tier for the Rhizome connector?

Yes. Rhizome's FAQ says every customer who creates an account gets 10 messages a month free, and its llms.txt says the MCP server has a free tier with OAuth sign-in. The free plan covers FDA, EMA and other English-language health authorities on the standard model. Paid plans remove the message limit, and REST API access is separately limited to the Enterprise plan.

Which health authorities does Rhizome cover?

FDA, EMA, MHRA, Health Canada, Swissmedic, TGA and more, plus ICH, ISO and IMDRF, per Rhizome's FAQ. It lists 19 countries, 7 international organizations and 137 databases. Several jurisdictions sit on the Enterprise tier only, including Japan, China, South Korea, Brazil and India, as do FDA adverse-event reports and Health Canada's released clinical study reports.

Can other Rhizome users see my questions or documents?

No, according to Rhizome's FAQ. It says isolation is enforced in the database itself, with row-level security scoping every query to your account. Rhizome also states that each agent run's permissions are resolved from your identity before the model is invoked. Conversations you start from Claude appear in your Rhizome account, so the same account isolation applies to them.

Sources

Use it in an agent

Put Rhizome AI to work.

Connect Rhizome AI once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.

Call (650) 285-1019, Agentman.Our AI receptionist answers.