Model Context Protocol (MCP)

Consumer Intelligence & VoC analytics, inside your AI agent.

Pivony is a consumer intelligence platform with a native MCP server. CX leaders and VoC analysts can ask AI agents real questions about live customer feedback — without exporting CSVs or rebuilding context every Monday.

Read the CX leader guide: Why VoC platforms need MCP →

Why consumer intelligence platforms need MCP

Dashboards answer "what happened." AI agents answer "what should I do next?" MCP is the bridge — a standard protocol so agents can call your VoC platform safely, with org-scoped permissions.

AI Agent

Cursor, Claude, ChatGPT…

MCP Server

pivony-mcp (stdio / SSE)

Secure JWT

MCP key → 1-hour token

Pivony API

VoC, dashboards, Advisor

Ask VoC in natural language

“Which F&B themes spiked negative this week?” or “Summarize Advisor findings for the board deck.” Agents call Pivony tools instead of hallucinating from stale exports.

Close the insight-to-action loop

MCP lets agents pull widget KPIs, list dashboards, and continue Advisor sessions — so analysis stays tied to your org's live data model, not a one-off chat paste.

Enterprise-grade access control

Scoped MCP keys (read:workspace, read:dashboards, chat:advisor). Keys exchange for short-lived JWTs. Every tool call is audit-logged on the Pivony API.

Available MCP tools

Six tools ship today. Each maps to a Pivony API endpoint your CX team already uses in the console.

ToolScope
pivony_list_workspace_groupsread:workspace
pivony_get_widget_dataread:workspace
pivony_list_dashboardsread:dashboards
pivony_advisor_chatchat:advisor
pivony_list_advisor_sessionschat:advisor
pivony_get_user_contextread:org

Works with the AI tools your team already uses

Install pivony-mcp, paste your MCP key from Pivony Settings → Integrations → MCP (AI Agents), and connect Cursor, Claude Desktop, or a custom agent stack.

CursorClaude DesktopChatGPTCustom agents

Setup in five steps

1

Create an MCP key

In Pivony: Settings → Integrations → MCP (AI Agents). Copy the key — shown once.

2

Install pivony-mcp

pip install pivony-mcp (or contact support@pivony.com for install options).

3

Configure your agent

Add the MCP server JSON to Cursor Settings → MCP or Claude Desktop config.

4

Validate connection

Ask: “List my Pivony workspace groups.” The server exchanges your key for a 1-hour JWT.

5

Run VoC workflows

Query widgets, dashboards, and Advisor — grounded in consumer intelligence, not generic web training data.

MCP vs. generic ChatGPT for VoC

CapabilityPivony MCPGeneric LLM
Live VoC / dashboard dataYes — scoped API toolsNo — unless you paste exports
Org-specific topic modelYes — your Pivony taxonomyNo
Advisor on your feedbackYes — pivony_advisor_chatGeneric answers
Audit trailYes — API audit logsNo
Revocable accessYes — per-key scopesN/A

Security & governance

MCP keys are scoped per user and organization
Keys validate via POST /mcp/validate → 1-hour JWT
Revoked keys stop working immediately
Production API: https://app.pivony.com/api/v1
SOC-aligned infrastructure; data encrypted in transit and at rest

Give your CX team an AI agent that actually knows your customers

Pivony MCP connects consumer intelligence and voice-of-customer analytics to the AI tools CX leaders use every day.