Table of Contents

Configuring AI

This guide walks you through setting up AI providers in Semantic Modeler. AI powers three features: Model Documentation, the Model Health AI scan, and the DAX Debugger.

All configuration happens in Settings > AI.

AI Tasks

The settings page splits AI into two tasks, each with its own provider:

Task What It Powers Model Recommendation
Documentation & Analysis Model documentation and the Model Health AI scan A reasoning or general-purpose model (e.g., Claude, GPT-4o)
DAX Assistant Generating, explaining, and fixing DAX A code-oriented model (e.g., Codellama, Claude, GPT-4o)

You can use the same provider for both, or mix and match - for example, Claude through Anthropic for Documentation & Analysis and a local Codellama instance through Ollama for the DAX Assistant.

Providers

Each task supports the following providers:

Provider What It Is
FactFoundry Hosted (license) AI hosted by FactFoundry - no API key or configuration needed. Available with an AI-enabled license.
Anthropic (your API key) Connect directly to Anthropic's API using your own key.
OpenAI / Local (OpenAI-compatible) Any endpoint that speaks the OpenAI API format - including OpenAI itself, Ollama, LM Studio, and other local runtimes.
Azure OpenAI (your key) An Azure OpenAI resource you manage.
Azure AI Foundry - any model (your key) Non-Claude models (GPT, Llama, Mistral, Phi, etc.) deployed in your Azure AI Foundry resource.
Azure AI Foundry - Claude (your key) Claude models deployed through Azure AI Foundry. Uses a separate authentication flow from other Foundry models.

The DAX Assistant also has a Same as Documentation & Analysis option that mirrors whatever you configured for the first task.

Setting Up a Provider

Open Settings > AI, then configure each task section.

FactFoundry Hosted

  1. Select FactFoundry Hosted (license) from the Provider dropdown.
  2. No further configuration is needed - this uses your license automatically.

Anthropic

  1. Select Anthropic (your API key) from the Provider dropdown.
  2. Paste your Anthropic API key and click Save.
  3. Optionally select a model from the dropdown (or type one in). Leave it blank to use the default.
  4. Click Test connection to verify.

OpenAI / Local (OpenAI-compatible)

This provider works with OpenAI's API, Ollama, LM Studio, and any other runtime that serves an OpenAI-compatible endpoint.

  1. Select OpenAI / Local (OpenAI-compatible) from the Provider dropdown.
  2. Choose an Endpoint preset or select Custom to enter your own URL.
  3. Enter the Base URL. This must end in /v1. Examples:
    • OpenAI: https://api.openai.com/v1
    • Ollama: http://localhost:11434/v1
    • LM Studio: http://localhost:1234/v1
    • Custom endpoint: https://ai.myworknetwork.xyz/api
  4. Select or type a Model. For Ollama, use the exact tag (e.g., codellama:latest, qwen2.5-coder:latest).
  5. If your endpoint requires an API key, paste it and click Save. Local runtimes like Ollama and LM Studio typically don't need one.
  6. Click Test connection to verify.

Azure OpenAI

  1. Select Azure OpenAI (your key) from the Provider dropdown.
  2. Enter your Azure OpenAI Endpoint (e.g., https://my-resource.openai.azure.com/).
  3. Enter the Deployment name - this is your deployment name, not the base model ID (e.g., gpt-4o).
  4. Paste your API Key and click Save.
  5. Click Test connection to verify.

Azure AI Foundry - Any Model

Use this for GPT, Llama, Mistral, Phi, and other non-Claude models deployed in your Foundry resource.

  1. Select Azure AI Foundry - any model (your key) from the Provider dropdown.
  2. Enter your Foundry Models Endpoint (e.g., https://my-resource.services.ai.azure.com/models).
  3. Enter the Model / deployment name (e.g., gpt-4o, Llama-3.3-70B-Instruct).
  4. Paste your API Key and click Save.
  5. Click Test connection to verify.
Note

For Claude models deployed in Azure AI Foundry, use the Azure AI Foundry - Claude option instead.

Azure AI Foundry - Claude

  1. Select Azure AI Foundry - Claude (your key) from the Provider dropdown.
  2. Enter your Foundry Resource Name - this is the <name> portion of <name>.services.ai.azure.com.
  3. Paste your Foundry API Key and click Save.
  4. Optionally enter a Model / deployment name. Leave it blank to use the default.
  5. Click Test connection to verify.

Using Different Providers Per Task

You might want a large reasoning model for Documentation & Analysis (which benefits from understanding business context) and a fast code model for the DAX Assistant (which benefits from code-generation speed).

To set this up, configure Documentation & Analysis with one provider, then configure the DAX Assistant separately with a different provider. If you want both tasks to use the same settings, select Same as Documentation & Analysis in the DAX Assistant section.

Privacy

API keys are stored only on your machine and are sent directly to the provider you selected - never to FactFoundry. The same applies to the data sent during AI requests: only schema metadata (table names, column names, data types, DAX expressions, and relationship definitions) is sent. No raw data from your tables is ever transmitted.

Edit this page