> This page is for version v8.3.0 (default). > For other versions, use one of these documentation indexes: > - v8.3.0 (default): https://docs.workflows.com/v-8-3-0/llms.txt > For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.workflows.com/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.workflows.com/_mcp/server. # MCP Quick Start ![Mind Reasoner MCP](/_fern-img/466e07efd0202b28f29ec8dda3be88918c85104c2652efceb8ff717127f861ca.webp) > **Instant Access to Digital Twins** > > The MCP Server brings Mind Reasoner into your AI conversations. Create > > digital minds > > , upload transcripts, and run > > simulations > > without leaving Claude—no code required. ## What You'll Build In this guide, you'll: * Connect Claude to Mind Reasoner's MCP Server * Create your first digital mind * Upload conversation transcripts * Run your first simulation **Time required:** 5-10 minutes --- ## Prerequisites #### Get Your API Key Log in to [Mind Reasoner](https://app.workflows.com) and copy your API key from the dashboard. Keep it secure. #### Install Claude CLI If you haven't already, ensure Claude CLI is installed and you're logged in. #### Prepare a Transcript Have a conversation transcript file ready (.vtt, .pdf, or .docx format). This will be used to train your first mind . --- ## Setup (2 minutes) Connect Claude to the Mind Reasoner MCP Server: ```bash claude mcp add --transport http mind-reasoner \ https://mcp.mindsim.com/mcp \ --header "Authorization: Bearer {{YOUR_API_KEY}}" ``` **Restart Claude Desktop** for the connection to take effect. > **Connected** > > Mind Reasoner tools are now available in Claude. You can verify by asking: "What Mind Reasoner tools do you have access to?" --- ## Your First Mind (3 steps) Now let's create a digital mind and run your first simulation: #### Create and Upload **You:** ``` "Create a new mind for my sales manager named Alex, then upload the transcript at /Users/me/Documents/sales-calls.vtt" ``` **Mind Reasoner:** * ✓ Creates the mind * ✓ Gets a secure upload URL * ✓ Uploads your transcript file * ✓ Confirms when complete #### Wait for Training **You:** ``` "Let me know when Alex's mind is ready." ``` **Mind Reasoner:** Monitors the training status automatically and notifies you when complete (typically 5-15 minutes). While waiting, you can continue other tasks in Claude Code. #### Run Your First Simulation **You:** ``` "How would Alex handle a prospect who says our product is too expensive compared to competitors?" ``` **Mind Reasoner:** Runs a simulation and returns Alex's predicted response, including communication style, tone, and approach. > **You're Running Simulations** > > You've created a digital mind and run your first simulation. You can now ask unlimited questions about how Alex would handle any sales scenario. --- ## How It Works Behind the scenes, Claude Code uses Mind Reasoner's MCP tools to interact with the Mind Reasoner service: 🧠 ### Mind Creation Creates a digital entity to represent the person. This establishes identity and metadata. 📤 ### Data Upload Securely uploads transcripts using signed URLs. Your data goes directly to secure storage. 🤖 ### AI Training Analyzes conversation patterns to build a snapshot . Captures style, tone, and decision-making patterns. ⏳ ### Status Monitoring Automatically checks when training completes. No manual polling needed—Mind Reasoner handles it. ✨ ### Simulations Generates predictions based on trained patterns. Instant responses, unlimited scenarios. 🔄 ### Continuous Use Run unlimited simulations without re-training. Update with new data when patterns evolve. > **Natural Language Interface** > > You don't need to know technical details. Just describe what you want in plain English—Claude Code selects the appropriate tools, and Mind Reasoner executes them automatically. --- ## Example Conversations Here are natural ways to interact with Mind Reasoner through Claude Code: ### Discovering Existing Minds ``` You: "Show me all the minds I've created." Mind Reasoner: [Uses get_all_minds] "You have 3 minds: 1. Sarah - Customer Service Manager (ready for simulations) 2. Mike - Sales Executive (training in progress) 3. Jordan - Technical Support Lead (ready for simulations)" You: "Find minds related to sales." Mind Reasoner: [Uses search_minds] "Found 1 mind: Mike - Sales Executive" ``` ### Creating Multiple Minds ``` You: "Do I already have a mind for customer service?" Mind Reasoner: [Searches first] "Yes, you have 'Sarah - Customer Service Manager' already. Would you like to use that or create a new one?" You: "Create a new one called 'Customer Service Team Lead'." Mind Reasoner: [Creates new mind] "Created. Ready to upload transcripts." ``` ### Comparing Responses ``` You: "For both Sarah and Mike, show how they would respond to an angry customer whose order was delayed by 2 weeks." Mind Reasoner: [Runs simulations for both minds] "Here's how each would respond: Sarah (Customer Service): [Empathetic, solution-focused response] Mike (Sales): [Relationship-preservation, future-focused response]" ``` ### Batch Testing ``` You: "Test how our support team mind would handle these five scenarios: 1. Password reset issues 2. Billing disputes 3. Feature requests 4. Technical bugs 5. Account cancellations" Mind Reasoner: [Runs five simulations] "I've tested all five scenarios. I notice the responses for billing disputes are less consistent than other categories—this might indicate a training gap." ``` --- ## Common Tasks #### Sales Training ### Train New Reps **Scenario:** Your top closer's techniques need to scale across the team. **What to do:** ``` 1. "Create a mind for Mike, our top sales rep" 2. "Upload his sales call transcripts from the last quarter" 3. "Generate 10 common objections and show how Mike handles each one" 4. Share results with new reps for study ``` **Result:** New reps learn from your best performer 24/7. #### Quality Assurance ### Test Before Launch **Scenario:** New product launching next week—will support handle customer questions consistently? **What to do:** ``` 1. "Create a mind representing our support team using last month's chat transcripts" 2. "Test how they'd handle these 8 common questions about the new product launch" 3. Review responses for consistency 4. "I found gaps in questions 3 and 5. Let's create training materials for those topics." ``` **Result:** Identify and fix gaps before customers experience them. #### Executive Prep ### Anticipate Questions **Scenario:** Presenting proposal to CEO tomorrow—what questions will they ask? **What to do:** ``` 1. "Create a mind for our CEO using board meeting transcripts" (with consent!) 2. "Here's my proposal summary. What questions and concerns would they likely have?" 3. Review predicted questions 4. "Help me prepare responses to these three concerns" ``` **Result:** Enter meetings fully prepared with pre-emptive answers. --- ## Tips for Success 📊 ### Upload Quality Data * 30-50 conversations minimum * Recent transcripts (last 6 months) * Clean, clear speaker attribution * Diverse scenarios represented 🎯 ### Be Specific in Scenarios * Include emotional context * Mention specific constraints * Provide relevant background * State desired outcomes 🔄 ### Iterate and Refine * Run multiple scenarios * Look for patterns in responses * Check confidence scores * Update with new data quarterly --- ## Troubleshooting ### Mind Reasoner Tools Not Available **Solution:** 1. Verify server is registered: `claude mcp list` 2. Restart Claude Desktop or Claude Code 3. Ask: "What MCP tools do you have access to?" ### Authentication Errors **Solution:** * Double-check API key in Mind Reasoner dashboard * Ensure format: `Bearer YOUR_KEY` (note the space) * Try generating a new API key ### Snapshot Processing is Slow **Normal times:** * Small files (\< 1 MB): 3-5 minutes * Medium files (1-10 MB): 5-15 minutes * Large files (10-50 MB): 15-30 minutes **If slower:** * Check the snapshot status for errors * Verify file uploaded successfully * Check Mind Reasoner dashboard --- ## What's Next? 🧠 ### Learn Core Concepts Understand Minds, Snapshots, and Simulations: [Core Concepts](/core-concepts) 📖 ### See More Examples Explore real-world applications: [Workflow Examples](/workflows) 🔧 ### Technical Details Deep dive into MCP tools: [Tools Reference](/mcp-tools) > **Need Help?** > > Questions? Check the [Installation Guide](/mcp-installation) for detailed setup instructions or email [support@reasoner.com](mailto:support@reasoner.com).