AI Platform

Configure which AI model powers Prerak's features, and how to connect ChatGPT or Claude Desktop.

How AI Is Used in Prerak

Choosing an AI Model

Workspace admins can configure which AI provider and model to use.

1. Go to Settings → AI Platform
2. Under Provider, choose:
   • OpenAI (GPT-4o, GPT-4 Turbo)
   • Anthropic (Claude Sonnet 4.6, Claude Opus 4.6)
   • Local / Custom (Ollama, LM Studio, or any OpenAI-compatible endpoint)
3. Enter your API key for the chosen provider
4. Select the model from the dropdown
5. Click Save

You can set different models for different features (e.g., fast/cheap model for suggestions, powerful model for agents).

Reasoning Mode

For complex tasks (agents, analysis), you can enable extended reasoning:

1. Go to Settings → AI Platform → Advanced
2. Toggle Reasoning Mode for supported models (Claude Sonnet 4.6, Claude Opus 4.6, or OpenAI o1/o3)
3. Set a reasoning budget (token limit for thinking)

Reasoning mode is slower and costs more tokens but produces significantly better results for multi-step logic.

Connecting ChatGPT

Use ChatGPT to talk to your Prerak CRM in natural language.

1. Get your API key: Settings → API Keys → Create Key (read-write scope)
2. In ChatGPT, open the GPT store and search for Prerak
3. Add the Prerak GPT
4. In the setup screen, paste your API key
5. Start chatting:

   "How many deals do I have closing this month?" "Create a task for me to follow up with Jane Chen tomorrow"

Connecting Claude Desktop

1. Get your API key: Settings → API Keys → Create Key
2. Open Claude Desktop → Settings → MCP Servers
3. Click Add Server and enter:

   {
     "name": "Prerak",
     "url": "https://mcp.prerak.io",
     "apiKey": "pk_live_your_key_here"
   }
   

4. Restart Claude Desktop
5. Start a new conversation — Claude will now have access to your CRM

Local AI Server (Privacy-First Option)

For data-sensitive workspaces, you can run a local LLM:

1. Install Ollama or LM Studio on your machine or server
2. Pull a model: ollama pull llama3
3. In Prerak: Settings → AI Platform → Provider → Local / Custom
4. Enter the base URL: http://localhost:11434 (Ollama default)
5. Select the model name
6. Click Save

With a local server, AI data never leaves your network. Contact your workspace admin or Prerak support for detailed local server setup assistance.

AI Credit Usage

AI usage is metered based on tokens consumed (not API calls).

• View current usage: Settings → Billing → AI Usage
• Each plan includes a monthly token allowance
• Exceeding the allowance: either upgrade, or AI features pause until the next billing cycle

Troubleshooting

AI suggestions not appearing:

• Check that an API key is saved in Settings → AI Platform
• Verify the API key is valid (test it in the provider's dashboard)
• Check billing — you may have exceeded your token limit

Slow responses from agents:

• Reasoning mode significantly increases response time
• Try switching to a faster model for non-critical tasks

Local server not connecting:

• Ensure Ollama/LM Studio is running (ollama serve)
• Check the base URL includes port: http://localhost:11434
• Ensure your firewall allows Prerak server to reach the local URL