> ## Documentation Index > Fetch the complete documentation index at: https://docs.lightdash.com/llms.txt > Use this file to discover all available pages before exploring further. # Start Deep Research run > Start a durable Deep Research investigation. ## OpenAPI ````yaml https://raw.githubusercontent.com/lightdash/lightdash/refs/heads/main/packages/backend/src/generated/swagger.json post /api/v1/ee/projects/{projectUuid}/ai-deep-research openapi: 3.0.0 info: title: Lightdash API version: 0.3374.1 description: > Open API documentation for all public Lightdash API endpoints. # Authentication Before you get started, you might need to create a Personal Access Token to authenticate via the API. You can create a token by following this guide: https://docs.lightdash.com/references/personal_tokens license: name: MIT contact: name: Lightdash Support email: support@lightdash.com url: https://docs.lightdash.com/help-and-contact/contact/contact_info/ servers: - url: / security: [] tags: - name: My Account description: These routes allow users to manage their own user account. - name: Organizations description: >- Each user is a member of a single organization. These routes allow users to manage their organization. Most actions are only available to admin users. - name: Projects description: >- Projects belong to a single organization. These routes allow users to manage their projects, browse content, and execute queries. Users inside an organization might have access to a project from an organization-level role or they might be granted access to a project directly. - name: Spaces description: >- Spaces allow you to organize charts and dashboards within a project. They also allow granular access to content by allowing you to create private spaces, which are only accessible to the creator and admins. - name: Roles & Permissions description: >- These routes allow users to manage roles and permissions for their organization. externalDocs: url: https://docs.lightdash.com/references/roles - name: Query description: >- These routes allow users to execute and manage queries against their data warehouse. This includes metric queries, SQL queries, and retrieving query results. paths: /api/v1/ee/projects/{projectUuid}/ai-deep-research: post: summary: Start Deep Research run description: Start a durable Deep Research investigation. operationId: createAiDeepResearchRun parameters: - in: path name: projectUuid required: true schema: $ref: '#/components/schemas/UUID' requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/AiDeepResearchRequestBody' responses: '202': description: Accepted content: application/json: schema: $ref: '#/components/schemas/ApiAiDeepResearchRunResponse' default: description: Error content: application/json: schema: $ref: '#/components/schemas/ApiErrorPayload' security: [] components: schemas: UUID: type: string format: uuid description: |- Stringified UUIDv4. See [RFC 4112](https://tools.ietf.org/html/rfc4122) pattern: >- [0-9A-Fa-f]{8}-[0-9A-Fa-f]{4}-4[0-9A-Fa-f]{3}-[89ABab][0-9A-Fa-f]{3}-[0-9A-Fa-f]{12} AiDeepResearchRequestBody: properties: effort: $ref: '#/components/schemas/AiDeepResearchEffort' description: Server-owned execution budget tier. Defaults to medium. prompt: type: string required: - prompt type: object ApiAiDeepResearchRunResponse: $ref: '#/components/schemas/ApiSuccess_AiDeepResearchRun_' ApiErrorPayload: properties: error: properties: data: $ref: '#/components/schemas/AnyType' description: Optional data containing details of the error message: type: string description: A friendly message summarising the error name: type: string description: Unique name for the type of error statusCode: type: number format: integer description: HTTP status code required: - name - statusCode type: object status: type: string enum: - error nullable: false required: - error - status type: object description: |- The Error object is returned from the api any time there is an error. The message contains AiDeepResearchEffort: type: string enum: - high - medium - low - xhigh ApiSuccess_AiDeepResearchRun_: properties: results: $ref: '#/components/schemas/AiDeepResearchRun' status: type: string enum: - ok nullable: false required: - results - status type: object AnyType: description: |- This AnyType is an alias for any The goal is to make it easier to identify any type in the codebase without having to eslint-disable all the time These are only used on legacy `any` types, don't use it for new types. This is added on a separate file to avoid circular dependencies. AiDeepResearchRun: properties: completedAt: type: string nullable: true startedAt: type: string nullable: true updatedAt: type: string createdAt: type: string cancellationRequestedAt: type: string nullable: true errorMessage: type: string nullable: true budget: $ref: '#/components/schemas/AiDeepResearchBudget' result: allOf: - $ref: '#/components/schemas/AiDeepResearchReport' nullable: true status: $ref: '#/components/schemas/AiDeepResearchRunStatus' projectUuid: type: string aiDeepResearchRunUuid: type: string required: - completedAt - startedAt - updatedAt - createdAt - cancellationRequestedAt - errorMessage - budget - result - status - projectUuid - aiDeepResearchRunUuid type: object AiDeepResearchBudget: properties: maxResultRows: type: number format: double maxWarehouseQueries: type: number format: double maxToolCalls: type: number format: double maxTokens: type: number format: double maxRuntimeMs: type: number format: double required: - maxResultRows - maxWarehouseQueries - maxToolCalls - maxTokens - maxRuntimeMs type: object AiDeepResearchReport: properties: nextSteps: items: type: string type: array unresolvedQuestions: items: type: string type: array scope: type: string caveats: items: type: string type: array findings: items: $ref: '#/components/schemas/AiDeepResearchFinding' type: array summary: type: string required: - nextSteps - unresolvedQuestions - scope - caveats - findings - summary type: object AiDeepResearchRunStatus: type: string enum: - queued - running - failed - completed - partially_completed - cancelled AiDeepResearchFinding: properties: evidence: items: $ref: '#/components/schemas/AiDeepResearchEvidence' type: array confidence: $ref: '#/components/schemas/AiDeepResearchConfidence' summary: type: string title: type: string required: - evidence - confidence - summary - title type: object AiDeepResearchEvidence: properties: sourceUrl: type: string nullable: true sourceLabel: type: string sourceType: type: string enum: - lightdash - warehouse - web description: type: string title: type: string required: - sourceUrl - sourceLabel - sourceType - description - title type: object AiDeepResearchConfidence: type: string enum: - low - medium - high ````