Getting Started

FlautoPsy monitors your AI workflows for drift in real-time. Get alerts when your LLM prompts, outputs, latency, or cost deviate from normal behavior.

What is FlautoPsy?

FlautoPsy automatically detects when your AI workflows behave differently than expected. It learns normal behavior from your first 10 traces, then alerts you via Slack when anything changes significantly.

This is essential for production AI workflows running on Make.com, n8n, Zapier, or custom APIs. For more detail, see the marketing overview.

Quick Start in 5 Minutes

1. Sign Up

Create a free account at https://flautopsy.stralocroft.com/signup

2. Name Your Workspace

Choose a workspace name (e.g., "Production" or "Email Classification"). This groups all your monitoring data.

3. Get Your Webhook URL

Go to /setup and copy your unique webhook URL. This is where you'll send traces from your workflow.

4. Paste into Your Workflow

  • Make.com: Add an HTTP module, paste the webhook URL, and map your workflow outputs
  • n8n: Add an HTTP Request node, paste the webhook URL
  • Zapier: Use Webhooks by Zapier, paste the webhook URL
  • Custom: Use any HTTP client with the webhook URL

See integration guides for detailed step-by-step instructions.

5. Run Your Workflow 10+ Times

Send at least 10 traces to build a baseline. You can test manually or let your workflow run naturally.

6. Watch for Alerts

Once your baseline is mature (≥10 traces), drift alerts will appear in your workspace and (optionally) in Slack.

What You'll Need

  • A Make.com, n8n, or Zapier account with an existing AI workflow, OR
  • A custom application that can make HTTP requests
  • An active AI model (GPT-4, Claude, etc.) in your workflow
  • (Optional) A Slack workspace to receive drift alerts

Don't have a workflow yet? Create one or see concepts to understand what FlautoPsy monitors.

Next Steps