Tools Reference

Mind Reasoner provides eight MCP tools that work together in a logical sequence. You don’t need to memorize these tools—simply describe what you want in natural language, and Mind Reasoner will automatically select and execute the appropriate tools.

Available Tools

Integration Tools (Optional)

When you connect Gong, Fathom, or Fireflies to your MCP server, up to 9 additional tools become available alongside the core Mind Reasoner tools. These tools let you access existing transcript repositories without manual file uploads.

Gong Tools (3)

Fathom Tools (3)

Fireflies Tools (3)

Learn more:

• Integration Overview - What integrations are available and why use them
• Gong Setup - How to connect to Gong
• Fathom Setup - How to connect to Fathom
• Fireflies Setup - How to connect to Fireflies

Workflow

Discovery Phase (Optional)

Explore existing minds before creating new ones:

get_all_minds or search_minds
   ↓ Returns: List of minds with IDs

Setup Phase

Create and train new minds:

1. create_mind
   ↓ Returns: mindId, digitalTwinId
2. get_signed_upload_url
   ↓ Returns: signedUrl, artifactId
3. upload_file_to_signed_url
   ↓ Confirms upload
4. create_snapshot
   ↓ Returns: snapshotId
5. get_snapshot_status
   ↓ Waits for: status = "completed"
6. simulate
   ↓ Returns: AI predictions

Workflow Phases

Setup Phase

Create the digital entity and upload conversation data.

Training Phase

Process uploaded data into a trained AI model.

Simulation Phase

Run unlimited predictions using the trained model.

Common Patterns

Discovering Existing Minds

Check what minds already exist before creating new ones.

Time: Seconds

Natural language example:

"Show me all minds I've created"
"Find minds related to sales"
"Do I already have a mind for Sarah?"

First-Time Setup

Run the core workflow to create a mind, upload data, train the model, and run your first simulation.

Time: 10-20 minutes (mostly waiting for training)

Natural language example:

"Create a mind for my sales manager using the transcript
at /Documents/sales-calls.vtt, then simulate how they'd
handle pricing objections."

Adding New Data

Update an existing mind by skipping step 1 (mind already exists) and running steps 2-6 with new data.

Time: 7-17 minutes

Running Multiple Simulations

After training is complete, run simulate as many times as needed without re-training.

Time: Seconds per simulation

Perfect for:

• Testing multiple scenarios
• Training team members
• QA validation
• Batch predictions

Error Handling

All tools return standard HTTP error responses. Each tool’s documentation page includes detailed error handling guidance.

See Also