Connecting Pulse via MCP
Query your Pulse coverage directly from the AI tools your team already uses, with every answer traceable back to its source.
Please note: MCP access is available as a separate add-on to your Pulse subscription and is set up for you directly by the Pulse team.
Why use MCP
The Model Context Protocol (MCP) lets an AI client, including Claude, Claude Code, Cursor, or Gemini, query Pulse’s structured mining data directly in conversation rather than requiring you to export a table and paste it in.
Because the underlying data is structured and cited rather than scraped from a webpage, the assistant answers from real figures and can point back to the filing behind each one instead of inventing something plausible.
This is a different route into Pulse data from Pulse AI, the chat tool inside the platform itself under Tools → Pulse AI.
To learn how to use Pulse AI inside the platform:
→ Read: Using Pulse AI
Pulse AI lives inside Pulse; MCP is how your own AI tools, outside Pulse, reach the same data.
What you can do
Once connected, you can ask your AI client questions such as:
- “Which copper projects in Chile have a post-tax NPV above US$1bn, and what study is each figure from?”
- “Show me every asset where Newmont holds a royalty rather than an operating interest.”
- “Which ASX-listed explorers have under twelve months of cash runway?”
The assistant queries live Pulse data to answer, covering companies, assets, production, reserves, drilling, ownership, deals, people, documents, tenure, and financial health.
It can also cite the source behind each figure it returns.
How your data is handled
MCP is one-way and read-only, by design and by enforcement: query tools only, SELECT-only database access, with every query run in a read-only transaction.
There is no capability to create, edit, delete, or publish anything, and no standing database credentials sit on the MCP server.
Pulse never sees your conversation. Your AI client sends a query and Pulse returns the data; the reasoning and context around that query stay inside your own AI environment.
On the Pulse side, only the query itself - including its text and parameters - is logged for reliability and security purposes and is purged after 90 days. Results are not logged at all.
Pulse also performs no inference on the MCP path. The model doing the reasoning is the one you’re already using, running under your own agreement with that provider.
Infrastructure is dedicated per client rather than shared and is hosted on AWS in the EU (Stockholm) by default, with other regions available on request.
Data in transit and at rest is encrypted throughout.
Getting MCP access
MCP access is a paid add-on, separate from your platform seats, and is set up for you directly by the Pulse team.
Email sales@pulseintelligence.com to request pricing and access to MCP.
The Pulse team sets up the connection for your organisation.
You then connect your AI client to Pulse over OAuth, signing in with your own Pulse account.
Because the connection runs over OAuth against your individual account, the assistant only ever sees what you’re already permitted to see on the platform. Access follows you rather than a shared credential.
Actions you can take
With MCP access, you can:
- Request access via sales@pulseintelligence.com.
- Connect a supported AI client, including Claude, Claude Code, Cursor, or Gemini, to your Pulse account over OAuth.
- Ask questions in plain language and get answers grounded in your Pulse coverage, with sources cited.
- Point your security team here if they need the read-only, no-results-logging position before signing off on access.
- Use MCP alongside Pulse AI: MCP is for your own external tools, while Pulse AI is for working inside the platform itself.
Where to find full documentation
Setup instructions for each supported AI client are available in the API documentation at docs.pulseintelligence.com/docs/mcp.
Next steps
To learn more about accessing Pulse data directly through the API:
→ Read: Exploring API access