DeeCee.ai Agentic API
API Endpoint ReferenceDecisions

Create a new decision

Use this when a user says: 'help me decide whether to...' or 'I need to make a decision about...'. Creates a new decision brief that can then be developed via /interview and /analyze.

POST/v1/decisions

Authorization

AuthorizationBearer <token>

API key in dc_live_... / dc_test_... format, from /settings/api-keys.

In: header

Request Body

decision_titlestring

A short, clear title for the decision being made (5-10 words).

Length1 <= length <= 200
decision_context?string

The full background for this decision: what's being decided, why, and any constraints or stakeholders already known. Optional — if the context isn't known yet, omit this and use interview_next_question followed by synthesize_context to build it up conversationally instead.

Lengthlength <= 20000
framework_hint?string

Optional. If omitted, the caller should use list_frameworks/GET /v1/frameworks guidance to pick one, or default to DACI for simple tactical decisions.

Value in"DACI" | "SPADE" | "helix"
use_case_id?string

Optional. An id from list_use_cases/GET /v1/use-cases. When set, interview_next_question uses that use case's guiding questions instead of a generic prompt for a more targeted conversation. Some use cases require a Pro/Beta subscription.

Response Body

curl -X POST "https://api.deecee.ai/v1/decisions" \  -H "Content-Type: application/json" \  -d '{    "decision_title": "string"  }'
{
  "success": true,
  "data": {
    "decision_id": "string",
    "title": "string",
    "framework": "DACI",
    "status": "string",
    "created_at": "2019-08-24T14:15:22Z"
  }
}
{
  "success": false,
  "error": "string",
  "code": "string"
}
{
  "success": false,
  "error": "string",
  "code": "string"
}
{
  "success": false,
  "error": "string",
  "code": "string"
}
{
  "success": false,
  "error": "string",
  "code": "string"
}

Ask the AI to revise a completed stage (does not save)

Returns a job_id — poll GET /v1/jobs/{id} for original_content and revised_content. Does NOT modify the decision: show the revision to the user, then commit it (as-is or edited) via POST /v1/decisions/{id}/refine/apply.

Synthesize a decision context from the interview conversation

Call once interview_next_question has gathered enough signal. Synthesizes a coherent context from the conversation and saves it to the decision. Also attempts to classify the conversation against the existing use-case catalog for parity/analytics — falls back to a synthetic id if nothing matches well, without forcing the interview into that template's shape.