How does automatic data fetching work after I set up an import?

Once the AI import builder completes setup, there is no further AI involvement. All subsequent data fetching is handled by an automated scheduler using pure logic — no AI processing on every fetch.

How the scheduler works

The data service runs a background scheduler that checks every 60 seconds for connections that are due for a fetch. When your connection's "next fetch" time arrives, the scheduler:

1. Retrieves your stored API credentials (encrypted at rest)
2. Calls your vendor's API endpoint using the configured authentication
3. Parses the response using the field mappings created during setup
4. Converts raw data into meter readings (timestamp + value + meter ID)
5. Stores the readings in the database
6. Calculates the next fetch time based on your interval

Fetch intervals

You choose how often data is fetched during setup. Common options:

• Every 15 minutes — for near-real-time monitoring
• Every hour — standard for most energy meters
• Every day — for daily summary data
• Custom cron schedule — for specific timing requirements

What happens on errors

If a fetch fails (e.g., the vendor API is temporarily down), the system uses exponential backoff:

• 1st failure: retry after 5 minutes
• 2nd failure: retry after 10 minutes
• 3rd failure: retry after 20 minutes (capped at 1 hour) After the maximum retry count (default 3 consecutive failures), the connection status changes to "error" and stops fetching automatically. You will see a red status indicator in Analytics → Data Import.

To recover from errors

1. Go to Analytics → Data Import and find the connection with an error status
2. Check the error message (e.g., "401 Unauthorized" means credentials expired)
3. Fix the underlying issue (update credentials, contact vendor)
4. Re-enable the connection — it will resume fetching on the next interval

Where to monitor: Analytics → Data Import shows the status, last fetch time, and next scheduled fetch for each active connection.