Summarize Deals: Claude + Make

A modular-home builder runs a dozen project deals at once, each a tangle of permits, timelines, and financing. The CRM holds the data, but reading every deal to find the one quietly going sideways takes an afternoon nobody has, so problems surface late, when they are expensive.

This build keeps a scorecard. When a new deal enters the pipeline, Claude reads it, writes a plain-language summary, and flags the risk level, all logged to a tracking sheet. A glance down one column shows which deals are healthy and which need a call today.

We wire it in Make, the last build in our Sales and CRM set. We watch HubSpot deals here, but any CRM with a deals trigger works. Claude turns a record into a read; you decide where to spend your attention.

What does this workflow actually do?

In one line: a new deal becomes a logged summary with a risk flag. Make watches your CRM’s deals. When one is created, Claude writes a plain summary and rates the risk, and Make logs it to a sheet. Your pipeline becomes a scannable scorecard instead of a stack of records.

A few real cases, none of them the usual ones:

• The modular-home builder above, spotting the shaky deal before it slips.
• A boutique event-production company sizing each new event deal by complexity.
• A B2B translation agency triaging new project deals by scope and deadline.
• A solar-EPC firm tracking pipeline deals by permitting and financing risk.

The CRM captures the deal; that part is automatic. Reading each one to judge its health is the review that slips when you are busy. That read is what Claude does in seconds.

Why use Make if Claude can already read and write?

Reasonable question. Claude is good at reading a deal record and summarizing its state and risk. It is not built to watch the pipeline and log rows to a sheet. That is plumbing, and Make does plumbing.

Make handles the no-judgment steps: catching the new deal, writing the row, running on its own. Claude handles the one step that needs a read, the summary and risk call. Make is the ledger; Claude is the analyst who writes the one-line verdict. You still decide what to do, but you start from a sorted scorecard instead of raw records.

What do you need before you start?

• A Make account. The free 1,000 operations a month covers a busy pipeline.
• A Claude API key from the console, not the chat app. See how to use Claude.
• A CRM with a new-deal trigger. We use HubSpot; others work the same way.
• A Google Sheet with columns for the deal, a Summary, and a Risk flag.

Two Make words. A scenario is the whole deal-to-scorecard automation. A module is one box. Three boxes, one scenario.

How does the workflow work, step by step?

Three modules, left to right:

Module	App	What it does
1. Trigger	HubSpot (CRM)	Fires when a new deal is created
2. Brain	Claude	Summarizes the deal and rates the risk
3. Output	Google Sheets	Logs the summary and risk flag as a row

Step 1: Watch your pipeline for new deals

Create a scenario and add HubSpot CRM, Watch Deals. Connect your CRM and watch for new deals. This trigger hands the next step the deal’s details. Another CRM works the same way: swap this one module.

Step 2: Let Claude read the deal

Add a Claude module, Create a Prompt, paste your key, pick Sonnet:

Summarize this new deal for a pipeline scorecard.
Deal: {{fields from Step 1}}

Return one line, pipe-separated: one-sentence summary | risk
(low/med/high) | the single biggest risk in a few words. Judge risk
only from the deal details, not assumptions. If thin, say so.

Asking for the single biggest risk, not a paragraph, keeps the scorecard scannable and forces Claude to pick what matters.

Step 3: Log it to the scorecard

Add Google Sheets, Add a Row, and map the deal name plus Claude’s pipe-separated Summary, Risk, and biggest-risk note. Sort or color the Risk column, and your whole pipeline reads at a glance.

How do you run it and check the result?

Click Run once. Make takes a recent deal, Claude summarizes and rates it, and a row lands in your scorecard. When I tested this, the risk flags were sensible when the deal carried real detail, and Claude said so plainly when a record was too thin to judge, which is what you want instead of a confident guess. Tune the prompt, then turn the scenario on.

After that, your pipeline reads like a dashboard. You scan the Risk column, spend your day on the high-risk deals, and stop discovering problems a week too late.

What does this cost to run?

Piece	Free tier	If you outgrow it
Make	1,000 operations/month free	Core plan from about $9/month
Claude API	Pay per use	Reading one deal on Sonnet costs a fraction of a cent
HubSpot	Free CRM tier works	Paid tiers as you grow
Google Sheets	Free	Free

Each deal is three Make operations and a sliver of Claude. A busy pipeline still sits inside the free plan. More on tiers in our Make guide.

What can go wrong, and how do you avoid it?

• Every deal flags high. The records are thin, or the prompt is vague. Define what high risk means and feed real detail.
• It guesses risk with no basis. Tell Claude to judge only from the deal details and to say when a record is thin.
• The columns misalign. Tell Claude to return only the pipe-separated fields.
• Deal data is confidential. Lock down the sheet, and mind what you send to any external model.

The same read-and-log pattern runs many builds. See Make AI scenarios.

How do you build this in Zapier or n8n instead?

Same three jobs, different names.

Job	Make	Zapier	n8n
Catch a deal	Watch Deals	New Deal trigger	HubSpot Trigger node
Summarize it	Claude module	Claude (Anthropic) action	Anthropic node
Log the row	Add a Row	Create Spreadsheet Row	Google Sheets node

Make and Zapier are easiest to start. Zapier vs Make vs n8n compares all three.

Common questions

Can it watch a Sheet of deals instead of a CRM?

Yes. If your pipeline lives in a sheet, use Watch New Rows as the trigger. The Claude and logging steps are unchanged.

How does it judge risk?

By the criteria you put in the prompt, read against the deal details. Define what low, medium, and high mean for your business.

What if a deal record is thin?

The prompt tells Claude to say so rather than guess, so a sparse deal flags as uncertain instead of getting a false confident rating.

Can it alert me on high-risk deals?

Yes. Add a filter and a Slack or email step after Claude so only high-risk deals ping you, on top of logging them all.

Is the chat app the same as the API?

Same models, different door. Make needs an API key from the console, not the chat login.

Primary sources

• Creating a scenario (Make Help)
• Claude API overview (Anthropic)
• Google Sheets basics (Google)
• Claude on Make (integrations)
• Sales pipeline — Grokipedia

Last reviewed: May 2026. Make, Claude, HubSpot, and Google update their interfaces; check the official pages for exact button names.

You may also like

• AI Automation for Beginners (the hub)
• Enrich a lead list with Claude + Make
• Draft Gmail replies with Claude + Make
• Make.com Complete Beginner’s Guide
• How to use Claude AI
• Make AI Scenarios: automations that save hours
• The AI Tools Directory