AI-Powered Computer Control for Windows

LM Studio Quick Start - 5 Minutes to Local AI! 🚀

Install Quickstart Releases

LM Studio Quick Start - 5 Minutes to Local AI! 🚀

What You Get

  • ✅ Run AI locally on your machine
  • ✅ Complete privacy - no data sent to cloud
  • ✅ No API costs - completely free
  • ✅ Works offline after setup
  • ✅ Same features as Azure OpenAI

Step 1: Download LM Studio (2 min)

  1. Go to https://lmstudio.ai/
  2. Click Download for your OS (Windows/Mac/Linux)
  3. Install and launch LM Studio

Step 2: Get a Model (3 min)

In LM Studio:

  1. Click the 🔍 Search icon (top left)
  2. Search for: hermes-2-pro
  3. Find: NousResearch/Hermes-2-Pro-Mistral-7B-GGUF
  4. Click Download next to Q4_K_M version (~4GB)
  5. Wait for download to complete

Alternative Quick Models:

  • Fast & Lightweight: Phi-2-GGUF (~1.6GB) - Great for testing
  • Best Quality: Mixtral-8x7B-Instruct-GGUF (~26GB) - Needs 32GB+ RAM

Step 3: Start LM Studio Server (30 sec)

  1. In LM Studio, click 💻 Local Server tab (left sidebar)
  2. Select your downloaded model from the dropdown
  3. Click Start Server button
  4. Wait for “Server started on http://localhost:1234”
  5. Keep LM Studio running!

Step 4: Configure FlowVision (1 min)

Option A: Using the UI (Easiest)

  1. Open FlowVision
  2. Go to Settings → LM Studio Configuration
  3. Check ☑ “Enable LM Studio”
  4. Click “Test Connection” (should show ✓ success)
  5. Click “Save”

Option B: Manual Configuration

Create file: %APPDATA%\FlowVision\lmstudioconfig.json

{
  "EndpointURL": "http://localhost:1234/v1",
  "ModelName": "local-model",
  "APIKey": "lm-studio",
  "Enabled": true,
  "Temperature": 0.7,
  "MaxTokens": 2048,
  "TimeoutSeconds": 300
}

Step 5: Test It! (30 sec)

  1. In FlowVision, type: “What is 2+2?”
  2. You should see: “Local AI response” indicator
  3. Get a response from your local model!

✅ Success Indicators

You’ll know it’s working when you see:

  • ✅ “LM Studio Action Execution” in the task indicator
  • ✅ “Processing your request with local AI” message
  • ✅ “Local AI response” during generation
  • ✅ Responses coming without internet connection

🎯 Try These Commands

Simple Test

"Tell me a joke"

Tool Calling Test

"Open Notepad"
"What windows are open?"
"Create a file called test.txt with the content 'Hello World'"

Code Generation

"Write a Python function to calculate fibonacci numbers"

🐛 Troubleshooting

“Cannot connect to LM Studio”

Fix:

  1. Make sure LM Studio is running
  2. Check that server is started (green “Stop Server” button visible)
  3. Verify endpoint: http://localhost:1234/v1

Model is too slow

Fix:

  1. Try a smaller model (Phi-2)
  2. Enable GPU acceleration in LM Studio settings
  3. Close other applications
  4. Reduce Max Tokens to 1024

“Connection timeout”

Fix:

  1. First request is slower (model loading)
  2. Increase timeout in config (300 → 600 seconds)
  3. Wait a bit longer
  4. Check CPU/RAM usage

Test Connection shows error

Fix:

  1. Restart LM Studio server
  2. Reload model in LM Studio
  3. Check firewall isn’t blocking port 1234
  4. Try changing port in LM Studio server settings

📊 Performance Tips

For Faster Responses:

  • Temperature: 0.2 (more focused)
  • Max Tokens: 1024 (shorter responses)
  • Model: Use Phi-2 or TinyLlama

For Better Quality:

  • Temperature: 0.7-0.8 (more creative)
  • Max Tokens: 2048-4096 (longer responses)
  • Model: Use Mixtral or Llama-2-13B

For Tool Calling:

  • Best Model: Hermes-2-Pro-Mistral-7B
  • Temperature: 0.3-0.5
  • Max Tokens: 1024

🔄 Switching Between Local and Azure

Enable Local AI:

Check ☑ “Enable LM Studio” in settings

Disable (Use Azure):

Uncheck ☐ “Enable LM Studio” in settings

No restart needed! FlowVision automatically switches.

💡 Pro Tips

  1. Keep LM Studio running - Don’t close it while using FlowVision
  2. First request is slow - Model loading takes time, be patient
  3. Monitor resources - Check Task Manager for CPU/RAM usage
  4. Try different models - Each has different strengths
  5. Use GPU - Enable in LM Studio settings for huge speed boost

📱 Quick Reference

Setting Recommended Fast Quality
Model Hermes-2-Pro-7B Phi-2 Mixtral-8x7B
Temperature 0.5 0.2 0.7
Max Tokens 2048 1024 4096
RAM Needed 16GB 8GB 32GB+

🎉 That’s It!

You’re now running AI locally!

Next Steps:

🆘 Need Help?

  1. Check full guide: LM Studio Integration Guide
  2. LM Studio docs: https://lmstudio.ai/docs
  3. LM Studio Discord: https://discord.gg/lmstudio
  4. GitHub Issues: (your repo)

Enjoy your private, local AI! 🎊

No more API costs! No more cloud dependency! Complete control! 🚀