Weekday competitor pricing watcher with Browserbase and Slack

Build a code workflow that runs every weekday at 8am America/New_York and watches a configurable list of competitor pricing page URLs. The list of URLs (with a friendly competitor name and an optional product/tier label per row) should be a workflow input, or read from a small config tab in the same Google Sheet used for logging.

For each URL, run the following deterministic per-URL flow in parallel where possible:

1. Browserbase Create Session to provision a fresh headless browser session and get its connect URL.

2. Drive that session with Stagehand or Playwright over the connect URL. Navigate to the competitor URL, wait for the pricing content to render, then extract the current sticker price as a number, the plan or tier name, any "starting at" or "from" copy near it, and the effective currency. Capture a screenshot of the rendered page and keep its public URL.

3. Browserbase Release Session immediately after extraction so the session does not sit idle.

Before writing the new row, use Google Sheets Get Values to read the most recent prior snapshot for that competitor and product from the Pricing Log tab (or a separate Latest Snapshot tab), so you can compute the absolute and percent change.

Then use Google Sheets Append Values to write one row per competitor per run to the Pricing Log tab with these columns: timestamp (ISO), competitor name, product or tier, URL, price (numeric), currency, raw starting-at copy, screenshot URL, previous price, absolute change, percent change. Both raw and percent change columns must be present so the sheet doubles as a pricing dashboard.

For every row whose absolute percent change exceeds a configurable threshold (default plus or minus 3 percent), use Slack Send a Message to post an alert to a configurable channel that names the competitor and product, shows the old price, the new price, and the percent change, and includes both the screenshot URL and a link to the source pricing page. Suppress the alert when the change is below the threshold or when there is no prior snapshot (the first run for a new URL only logs a baseline).

Keep the Browserbase step deterministic: one session per URL, fixed Stagehand instructions or Playwright selectors per competitor, always capture the screenshot, always release the session. Handle per-URL failures gracefully so one broken page never blocks the rest of the run, and log a clear row in the sheet (or a separate Errors tab) when a URL fails so the team can fix it.