For the complete documentation index, see llms.txt. This page is also available as Markdown.

Kai Resolution Agent

Kai resolution agent session management and action execution

Start agent session

post

Initiates a new Kai resolution agent session for a signal. The agent will analyze the signal and prepare to propose resolution actions. Requires create_agent_remediation permission and agentAI plan feature.

Authorizations
connect.sidstringRequired

Session-based authentication via HTTP-only cookie

Path parameters
idintegerRequired

Signal ID to create session for

Responses
201

Agent session created

application/json
post/concierge/alerts/{id}/agent-session
POST /api/concierge/alerts/{id}/agent-session HTTP/1.1
Accept: */*
{
  "session": {
    "id": 1,
    "alertId": 1,
    "status": "started",
    "createdAt": "2026-01-01T00:00:00.000Z"
  }
}

Get agent session status

get

Returns the current status and details of an agent session. Requires view_agent_remediation permission.

Authorizations
connect.sidstringRequired

Session-based authentication via HTTP-only cookie

Path parameters
idintegerRequired

Agent session ID

Responses
200

Agent session details

application/json
idintegerRequired

Unique session identifier

alertIdintegerRequired

Associated signal ID

statusstring · enumRequired

Current session status

Possible values:
createdAtstring · date-timeRequired
errorstring · nullableOptional

Error message if session failed

updatedAtstring · date-timeOptional
get/concierge/agent-sessions/{id}
GET /api/concierge/agent-sessions/{id} HTTP/1.1
Accept: */*
{
  "id": 1,
  "alertId": 1,
  "status": "started",
  "createdAt": "2026-01-01T00:00:00.000Z",
  "analysis": {
    "rootCause": "text",
    "impact": "text",
    "affectedSystems": [
      "text"
    ],
    "entityLifecycleContext": "text",
    "isAnomalyPattern": true,
    "patternDescription": "text",
    "confidence": 1
  },
  "proposedActions": [
    {
      "id": "action_1",
      "actionType": "reprocess_usage_batch",
      "description": "text",
      "isAutomated": true,
      "targetSystem": "Stripe",
      "targetEntities": [
        "text"
      ],
      "expectedOutcome": "text",
      "risk": "low",
      "steps": [
        {
          "order": 1,
          "description": "text",
          "apiCall": "text"
        }
      ]
    }
  ],
  "executionResults": [
    {
      "actionId": "text",
      "success": true,
      "response": {},
      "error": "text",
      "executedAt": "2026-01-01T00:00:00.000Z",
      "duration": 1
    }
  ],
  "error": "text",
  "updatedAt": "2026-01-01T00:00:00.000Z"
}

Generate action proposals

post

Triggers Kai to analyze the signal and generate proposed resolution actions. The agent performs root cause analysis and identifies systemic issues rather than isolated failures. Actions are categorized as automated (Kai can execute) or manual (requires user intervention). Requires create_agent_remediation permission and agentAI plan feature.

Authorizations
connect.sidstringRequired

Session-based authentication via HTTP-only cookie

Path parameters
idintegerRequired

Agent session ID

Responses
200

Actions proposed successfully

application/json
post/concierge/agent-sessions/{id}/propose
POST /api/concierge/agent-sessions/{id}/propose HTTP/1.1
Accept: */*
{
  "session": {
    "id": 1,
    "status": "awaiting_confirmation",
    "alertId": 1,
    "analysis": {
      "rootCause": "text",
      "impact": "text",
      "affectedSystems": [
        "text"
      ],
      "entityLifecycleContext": "text",
      "isAnomalyPattern": true,
      "patternDescription": "text",
      "confidence": 1
    },
    "proposedActions": [
      {
        "id": "action_1",
        "actionType": "reprocess_usage_batch",
        "description": "text",
        "isAutomated": true,
        "targetSystem": "Stripe",
        "targetEntities": [
          "text"
        ],
        "expectedOutcome": "text",
        "risk": "low",
        "steps": [
          {
            "order": 1,
            "description": "text",
            "apiCall": "text"
          }
        ]
      }
    ]
  }
}

Execute approved actions

post

Executes the specified resolution actions that the user has approved. The agent will make API calls to the relevant billing/payment systems and report results. Requires execute_agent_remediation permission and agentAI plan feature.

Authorizations
connect.sidstringRequired

Session-based authentication via HTTP-only cookie

Path parameters
idintegerRequired

Agent session ID

Body
actionIdsstring[]Required

IDs of approved actions to execute

Example: ["action_1","action_2"]
Responses
200

Actions executed

application/json
post/concierge/agent-sessions/{id}/execute
POST /api/concierge/agent-sessions/{id}/execute HTTP/1.1
Content-Type: application/json
Accept: */*
Content-Length: 37

{
  "actionIds": [
    "action_1",
    "action_2"
  ]
}
{
  "results": [
    {
      "actionId": "text",
      "success": true,
      "response": {},
      "error": "text",
      "executedAt": "2026-01-01T00:00:00.000Z",
      "duration": 1
    }
  ],
  "session": {
    "id": 1,
    "alertId": 1,
    "status": "started",
    "createdAt": "2026-01-01T00:00:00.000Z",
    "analysis": {
      "rootCause": "text",
      "impact": "text",
      "affectedSystems": [
        "text"
      ],
      "entityLifecycleContext": "text",
      "isAnomalyPattern": true,
      "patternDescription": "text",
      "confidence": 1
    },
    "proposedActions": [
      {
        "id": "action_1",
        "actionType": "reprocess_usage_batch",
        "description": "text",
        "isAutomated": true,
        "targetSystem": "Stripe",
        "targetEntities": [
          "text"
        ],
        "expectedOutcome": "text",
        "risk": "low",
        "steps": [
          {
            "order": 1,
            "description": "text",
            "apiCall": "text"
          }
        ]
      }
    ],
    "executionResults": [
      {
        "actionId": "text",
        "success": true,
        "response": {},
        "error": "text",
        "executedAt": "2026-01-01T00:00:00.000Z",
        "duration": 1
      }
    ],
    "error": "text",
    "updatedAt": "2026-01-01T00:00:00.000Z"
  }
}

Cancel agent session

post

Cancels an active agent session. Requires create_agent_remediation permission.

Authorizations
connect.sidstringRequired

Session-based authentication via HTTP-only cookie

Path parameters
idintegerRequired

Agent session ID

Responses
200

Session cancelled successfully

application/json
messagestringOptional
post/concierge/agent-sessions/{id}/cancel
POST /api/concierge/agent-sessions/{id}/cancel HTTP/1.1
Accept: */*
{
  "message": "text"
}

Last updated

Was this helpful?