Daily competitor blog digest in Slack

Build me a daily competitor blog digest as an agent workflow. Trigger: cron, every weekday at 8:00am Eastern Time.

Inputs the agent needs at run time: a Google Sheets spreadsheet ID and sheet/tab name. The sheet has one row per competitor with these columns: Competitor Name, Site URL (their blog or newsroom), and Last Seen URL (the most recent article URL the agent has already digested for that competitor; may be blank on first run).

Step 1. Use Google Sheets "Get Values" to read the full competitor table from the sheet.

Step 2. For each competitor row, call Diffbot "Analyze URL (Auto-detect Page Type)" on the listed Site URL to pull the latest article links and metadata from the blog or newsroom index page. From the result, identify the article URLs published since the Last Seen URL marker (if Last Seen URL is blank, take only the single most recent article so the first run does not flood Slack). Preserve the order so the newest article on the index becomes the new Last Seen URL after the run.

Step 3. For each new article URL, call Diffbot "Extract Article" to get clean title, author, publish date, and body text.

Step 4. The agent groups findings by competitor and writes a tight Slack digest. Format: a short header line ("Competitor digest, [date]"), then one section per competitor that has new posts. Each section is the competitor name in bold, followed by one paragraph per new article containing the headline (linked to the source URL) and a 2 to 3 sentence "why it matters" take. The agent should reason about what is genuinely interesting (product launches, pricing, positioning shifts, leadership changes, customer wins) versus boilerplate (generic thought-leadership, hiring posts, recycled content) and lean toward signal. Skip competitors with nothing new. If no competitor has anything new, post a one-line "All quiet on the competitor front today" message.

Step 5. Post the digest to Slack using "Send a Message" to the configured channel (default suggestion: #competitive).

Step 6. For every competitor that had at least one new article this run, use Google Sheets "Update Values" to write the newest article URL back into that competitor's Last Seen URL cell, so tomorrow's run only picks up posts newer than today's.

Error handling: if Diffbot fails on a single site, log it, skip that competitor for this run, and continue with the rest. Do not update the Last Seen URL for a competitor whose extraction failed.