ReadingMinds MCP Server Is Now GA: Secure Customer Truth as a Tool Call, Not a Dashboard

Today we are announcing general availability of the ReadingMinds MCP Server: a secure, governed way for Claude, GPT, or custom agents to query customer truth as a tool call, not a dashboard. REST API and webhooks are on the roadmap, and will share the same evidence schema when they ship.

MCP is rapidly becoming the "USB-C" for agent workflows. But as it graduates into production infrastructure, it is also entering the enterprise threat model: connectors, gateways, and tool endpoints are now a real perimeter. We built ReadingMinds MCP accordingly: secure by default, not "ship now, harden later."

What You Get with ReadingMinds MCP GA

Evidence Packs, Not Vibes

Every response is structured around quote objects pulled directly from transcripts, tagged with sentiment scores, themes, and full provenance back to the source conversation. Your agents do not get summaries or vibes. They get citable evidence: the exact words your customers used, the sentiment signal attached to those words, and the thematic context that makes them actionable.

The MCP server exposes 12 tools via Streamable HTTP transport. Any MCP-compatible agent can connect, query studies, pull analysis, and retrieve conversation-level evidence.

Governed Tool Access

Security teams should not have to fight with engineering over agent permissions. ReadingMinds MCP ships with 7 granular scopes: studies:read, studies:write, analysis:read, analysis:write, conversations:read, distribution:read, and distribution:write.

Every API key is created with explicit scope lists. No defaults, no implicit access. Each tool call checks for the required scope before executing. Keys are rate-limited, and key creation and revocation are logged through the security service. This is least-privilege by design, so security teams can approve deployments without drama.

Privacy-Minimizing by Default

No permanent voice recordings stored. We retain transcripts and derived sentiment signals. This is the same privacy posture we apply across the entire ReadingMinds platform: capture the insight, not the raw audio.

Why This Matters for RevOps and Customer Teams

Agents can now pull proof-grade customer language, and the sentiment behind it, before sending follow-ups, updating enablement, escalating churn risk, or rewriting messaging. Every quote traces back to a specific conversation, so the evidence holds up in a revenue meeting.

Consider what this enables in practice:

Retention workflows: An agent detects a negative sentiment cluster in recent interviews, pulls the exact quotes driving it, and drafts a save playbook with cited evidence before the CSM's next call.

Messaging optimization: A marketing agent queries the latest positioning study, extracts the phrases that scored highest in sentiment, and feeds them directly into ad copy or landing page tests.

Pipeline intelligence: A revenue agent pulls conversation evidence from recently stalled deals, identifies the specific objections and hesitation patterns, and surfaces them in the CRO's weekly review.

None of this requires a human to open a dashboard, export a CSV, or copy-paste quotes into a slide deck. The evidence flows directly into the workflow.

Under the Hood

For teams evaluating the technical architecture:

Transport: Streamable HTTP with API key authentication

Tools: 12 MCP tools covering studies, analysis, conversations, and distribution

Scopes: 7 granular permission scopes, least-privilege by default

Rate limiting: Per-key rate limits to prevent abuse

Key management: Auditable key creation and revocation via the security service

Privacy: No permanent voice recordings stored; transcripts and sentiment signals only

What Is Next

REST API and webhook support are in development. When they ship, they will use the same evidence schema, so integrations built on MCP today will translate cleanly. We are also working on emotion classification beyond the current sentiment scoring, and configurable data retention controls.

Get Started

If you are building agentic workflows and want your automation grounded in real customer truth, this is your integration surface. The MCP server is available now for all ReadingMinds customers on Launch, Pilot, and Enterprise plans.