Token per client
Each token resolves server-side to exactly one client account. The LLM never sees or passes that scope.
Connect Claude, Cursor, ChatGPT — any MCP-compatible LLM — to your live Reliz performance data. Per-client tokens. Strict per-client isolation. Always real-time.
Works with the AI tools your team already uses
MCP is the open standard your LLM uses to call out to live data. We exposed your Reliz performance data through it, so your team can ask their AI tool what they used to ask analysts. No SQL. No exporting CSVs. No dashboards.
Ask in natural language, get structured results. The LLM picks the right tool, fills in the parameters, and explains the answer.
See example queriesStandard MCP transport. Works with Claude, Cursor, VS Code, ChatGPT — and every future MCP-compatible client. No lock-in.
View compatibility listSix guarantees from the day you connect.
Every query hits the Reliz API directly. Always-current results, no waiting.
Tokens scope to one client account. Server-side enforced. You only ever see your own data.
Claude, Cursor, VS Code, ChatGPT — and every future MCP client. Standard transport.
Paste a JSON snippet with your token. The LLM picks up tools automatically.
No SQL, no dashboards. The LLM composes the right tool calls from your question.
Bearer-token auth, JSON config, standard HTTP/SSE transport. Same shape as every other MCP server.
From request to first answer, most teams are live in under a minute.
Email your account manager or hit the form. We issue a token scoped to your client account.
Paste the JSON snippet into your MCP client's config file. One block, two values.
Restart your LLM. Your team can now query Reliz data in plain English. No SQL.
{
"mcpServers": {
"reliz": {
"type": "http",
"url": "https://api.platform.reliz.com/mcp",
"headers": {
"Authorization": "Bearer <your-token>"
}
}
}
}Reliz MCP is a thin, audited bridge between your LLM and your live Reliz client data. Nothing trains, nothing persists outside your account.
Each token resolves server-side to exactly one client account. The LLM never sees or passes that scope.
One call from your account manager invalidates the token. Active sessions disconnect within seconds.
Queries flow through. Responses are not retained for model training or shared with other clients.
Every tool call is logged with timestamp, token prefix, and request ID. We surface anomalies to your account manager.
One token per client. Strictly scoped. Revocable on request.
Reliz MCP is in beta with selected clients. Request access and we'll get you set up in under a day.