September 9, 2026
A deal sitting in Contract Sent for sixteen days with a close date three days out. Nobody flagged it.
That’s not a sales problem. That’s a visibility problem. And if you’re not catching it before the week starts, you’re already behind.
We had seventeen open deals across multiple stages. Every week the question was the same: where do we focus first? For most teams, answering that means opening HubSpot, filtering by stage, checking last activity dates, and manually trying to figure out what’s at risk. That takes thirty minutes on a good Monday. On a busy one, it doesn’t happen at all.
We built a better way to run it.
Before any of this works, your HubSpot portal has to be connected inside Claude. Once it is, Claude has live access to your deals, stages, owners, and activity. No CSV export. No copy-paste. Just a direct line to what’s actually happening in your pipeline.
Without that connection, you’re back to doing this manually. With it, the whole workflow changes.
Skills in Claude are saved instruction sets that tell it how to run a specific workflow. You write them once and they’re ready every session.
The skill we built for pipeline review does four things every time it runs. It pulls all open deals from the connected HubSpot pipeline. It flags anything stalled over fourteen days. It assigns a risk level to each flagged deal. And it gives you one recommended next action per deal.
Once you’ve saved the skill, the whole review starts with one word: run.
Every flagged deal comes back in the same format. Risk level. Stall time. Deal amount. One next action.
No commentary. No filler. Just what the team needs to make a decision.
At the bottom of the review, deals that are clean show up under Deals Not Flagged. In a recent run, that was one deal out of seventeen. One.
The review that used to take thirty minutes now takes seconds. For more on why that kind of time loss compounds before anyone notices it, read Why Execution Capacity Is What’s Killing Your Pipeline. The drag is rarely one big thing. It’s the thirty minutes, every Monday, that never quite happens.
Every review ends with the same line: Review complete. Operator decision required before any action is taken.
That line is intentional.
Claude tells you what the data shows. It does not tell you what to do. The tool surfaces what’s at risk. The person decides what happens next. We built it that way on purpose.
There’s a version of AI-assisted RevOps where the system acts autonomously and the team finds out afterward. That’s not what we built. The operator stays in the seat. Claude makes sure they have the information they need to use it.
The skill is the same every time. What builds is the habit of running it.
You run it every Monday before the week starts. Nothing slips through because the review always happens. Deals don’t age in silence. Close dates don’t sneak up on the team. The pipeline reflects reality instead of assumptions.
That consistency is what makes it worth building. A one-time audit is useful. A repeatable system is what actually moves the number. If you want to see how the same logic applies to how we handle lead routing inside HubSpot, Your HubSpot Can Build Its Own Lead Routing Workflow. Most Teams Have No Idea. covers exactly that.
If your team is still doing this manually, scrolling records, checking dates, trying to remember who last touched what, try building the skill.
Connect HubSpot to Claude. Write the instruction once: pull open deals, flag anything stalled over fourteen days, assign a risk level, recommend one next action. Save it.
Next Monday, open Claude, run the skill, and you’ll have the answer before the coffee’s done.
If you want help building this out for your specific pipeline or want to see how we’d structure it for your stages and deal properties, book a call at trailblazer6.com.
