DocsWhat you can doKeep a table filled from a website

For agents

A table that fills itself.

Weather, prices, public data: LakehouseBox fetches it on a schedule and adds it to your table. Your computer can be off.

  • Agent-guided
  • Preview

Before you start: Claude Code (or any other agent)The website or API with the dataScheduled tables switched on for your account

Describe the data and how often

Say where the data comes from, what you want to keep and how often. Mention that your computer won't always be on.

Claude Code (or any other agent)Example conversation
You

I want a table that fills itself: every hour, the weather in Valencia from Open-Meteo (temperature, rain, wind), kept over time so I can look back. My computer won't always be on. Follow lakehousebox.com/docs/

Agent · example reply

The right tool here is LakehouseBox's Scheduled SQL: LakehouseBox runs the query on its own servers every hour and adds the rows to your table, so your computer doesn't need to be on.

Approve this computer in your browser (tick write). One more thing: scheduled tables are switched on per account by LakehouseBox. If it's off for you, you'll need to ask hello@lakehousebox.com.

Example replies show what to expect. Your agent adapts to your setup.

Your turn · in the browser

Approve, and check scheduled tables are on

Approve your agent's sign-in with Write ticked. Scheduled tables are a preview: if your agent reports "not enabled", write to hello@lakehousebox.com with your account's handle and we switch it on.

Your agent sets the schedule and runs it once

It writes the query, runs it once by hand to check the result, and tells you when it runs next.

Claude Code (or any other agent)Example conversation
You

Approved with write. Scheduled tables are on for us.

Agent · example reply

Your table is set up and will keep filling itself. LakehouseBox fetches the Valencia weather every hour at 10 minutes past, on its own servers.

  • Columns: time, temperature (°C), rain and precipitation (mm), wind speed and gusts (km/h), wind direction.
  • First run: I ran it once by hand: the last 7 hours were saved, the latest 22.6 °C, no rain, 3.2 km/h of wind.
  • Each run fetches the last 6 hours, so a missed run never leaves a gap; for one row per hour, ask me and I'll remove the repeats in the query.
Example replies show what to expect. Your agent adapts to your setup.

Example · verified connection

First run savedNext runs scheduledIn the recorded run: 7 hours saved by a manual run; hourly runs on LakehouseBox's servers after that.
What a scheduled table can read

Any public web address that returns data (JSON, CSV or Parquet): weather and air-quality services, open-data portals, price lists, your own exports published somewhere. For APIs that need a key, the key is stored once as a connection and never shown again. Each run has limits (time, requests and data downloaded) shown in your account's usage; runs can be paused, and each one is recorded, so you can see what was added and when.

  • Show me the warmest hour this week.

  • Add Madrid and Sevilla to the same table.

  • Pause the weather updates.