Your pipeline swept clean before Monday standup

What does the Monday pipeline sweep look like?

You describe the pass once: every open opportunity with a past close date, no next step, or an empty amount, grouped by owner. Push the close date where the owner logged activity this week, flag the rest for the owner to answer. The companion walks the pipeline, drafts the updates, and comes back with one batch. Run it as a scheduled routine and the report exists before anyone asks, which is the weekly pipeline review pattern.

Enrichment is the second pass. New signups on a free-mail domain with an empty company field get looked up one by one, cross-checked against Apollo or Crunchbase, both native, and written back as staged updates. The mechanics are in lead enrichment. Both jobs are one sentence each:

• “Every Monday at 7: list open opps with past close dates or missing amounts, stage the obvious fixes, flag the rest by owner.”
• “For each new signup with a company domain: fill industry, headcount and country from Apollo, and queue the writes for my review.”

Native for your CRM, session-driven for the rest

Native, meaning first-party tools rather than browser driving: Salesforce, Attio, Close, Twenty, Apollo, Crunchbase, Google Sheets and Slack. That covers CRM reads and writes, enrichment lookups, the forecast sheet, and the channel the digest lands in.

HubSpot is reached through the MCP server HubSpot publishes. Outreach, Gong and your billing admin get worked in your logged-in session: the companion opens the tab as you, reads what you could read, clicks what you could click. That is how the long tail of a revenue stack gets covered, including the internal admin panel nobody will ever build a connector for.

What RevOps keeps

Territory design, comp plans, forecast judgment, and the political work of getting sales to follow the process. A companion tells you stage-2 conversion dropped eight points after the pricing change. You decide whether that is noise, seasonality or a real problem. What it takes off you is the four hours of tab-switching that produced the number.

How does approval work when AI writes to the CRM?

Consequential actions pause on an approval card that shows exactly what is about to change, and Ask Always mode in Settings → AI permissions makes every write ask first. Batch edits arrive as reviewable batches, which matters in a system where a bad mass-update costs a week of cleanup. Finished runs of this work are public at /real-runs.

Frequently asked questions