ACC Get AEC Model AI

The ACC Get AEC Model AI node uses AI and GraphQL queries to extract specific element data from Autodesk Construction Cloud (ACC) AEC models based on user input. It combines OpenAI for natural language understanding with ACC APIs to return targeted BIM data.

Purpose

Use this node to intelligently interpret user queries (e.g. “Show me all the doors on Level 2”) and retrieve matching BIM model data automatically for reporting, dashboards, or downstream tasks.

Inputs

Field Label	Input Type	Description
Input Text	Text	User query describing the desired model data (e.g. “Show me all windows on Level 1”).
Mesh Record Id	Text	The Mesh Record ID from ElasticSearch containing the Revit data reference.
Return Properties	JSON Array	A list of property names to include in the returned data (e.g. ["Width","Height","Family Name"]).

Outputs

Output Variable	Description
AecData	The JSON string result from the AI-interpreted GraphQL query containing the requested AEC model data.

Functionality

This node:

1. Validates required inputs (Input Text, Mesh Record Id, Return Properties).
2. Retrieves service account tokens for OpenAI and Autodesk.
3. Fetches Revit data from ElasticSearch using the Mesh Record ID.
4. Uses AI classification to determine if the user query is level-based or general.
5. Formats family and category data from the model into a structured JSON.
6. Prompts AI to extract relevant groups and objects based on the user’s natural language query.
7. Builds a GraphQL query to retrieve model elements matching the interpreted request.
8. Executes the GraphQL query against the ACC AEC Data Model API.
9. Stores the returned data as AecData for use in subsequent workflow tasks.

Example Use Case

✅ User enters: “Show me all the doors and windows on Level 2.”
→ Node classifies as level-based, extracts relevant groups and objects, builds a filtered GraphQL query, and returns only doors and windows on Level 2.

✅ User enters: “List all mechanical equipment.”
→ Node returns all mechanical equipment elements with the specified properties.

Task Success Criteria

• Success: Relevant model data is retrieved from ACC and stored as AecData based on user query interpretation.
• Failure: If AI classification, ElasticSearch data retrieval, or GraphQL execution fails, the task fails with a descriptive error message indicating the failure point.

Notes

• Ensure ServiceAccount-OpenAI and ThirdParty-Autodesk service accounts are configured with valid credentials and appropriate API access.
• The Return Properties input must be a JSON array of property names to include in the output.
• The output (AecData) is a JSON string representing the GraphQLResponse from the AEC Data Model API.
• The AI prompts include advanced domain instructions to ensure accurate building object extraction while excluding irrelevant categories.