Your floor, queryable
by any agent.

The AI Agent API is a remote MCP server that exposes every metric, ticket, classifier, and conversation in Qvasa:

OAuth 2.1 with a scope picker at consent
Permissions follow the user, an agent never sees more than its human
Connect Claude, Gemini, or any MCP client in about six clicks

Request a demo Start free

Claude

Gemini

Any MCP client

// Orient
→ tools/call  qvasa_search_tools
  { "query": "chats waiting by brand" }
← 1 match   18 ms
  qvasa_query_metric  metrics:read

// Query
→ tools/call  qvasa_query_metric
  { "metric": "chats_waiting",
    "group_by": "brand",
    "window": "now" }
← 200       41 ms
  brand        waiting   longest
  Billing           12      3:42
  Shipping           4      0:58
  Returns            2      0:21

connected OAuth 2.1 scopes: metrics:read · tickets:read

Connect

Three install paths, and the fastest is about six clicks in Claude's settings. No token to copy.

Ask & build

Agents answer metric questions, build dashboards and calculated metrics, and create AI classifiers that backfill your history. They configure SLAs and alert routing too.

Teach

Runbooks are account-level playbooks any user's agent reuses. Attribute aliases teach the account your vocabulary, and errors ship recovery hints.

36 Permission scopes

7 Tools in the portal

3 Install paths

Put an agent
on the floor.

Start free, or see it on your own queues in a 30-minute demo.