Sales & CRM

How to run a weekly sales pipeline review with AI

Sales & CRM3 AI tools7 steps6 friction points

Want an agent to handle this on your live data? See the Starch version →

A weekly sales pipeline review is the meeting — or the solo ritual — where you go deal by deal, figure out what's stalled, what's close to closing, and where you need to apply pressure this week. For most operators running sales themselves or managing a small team, it's the difference between actually knowing your number and just hoping. The workflow sounds simple: look at your deals, update your forecast, decide what needs attention. In practice it eats 90 minutes and still leaves you unsure.

The reason people reach for AI is that the work is mostly analytical and repetitive in structure, even if the content changes week to week. You have a list of deals. You have activity data — emails, calls, last contact. You want a clear summary of what's at risk, what's on track, and what actions to take. That framing feels like exactly the kind of synthesis task where a language model should shine: take messy inputs, return structured insight.

ChatGPT, Claude, and Gemini can genuinely help here — particularly with synthesis and drafting. If you paste in a CRM export or a formatted deal list, a good model will spot patterns, flag stale deals, and help you write a crisp pipeline summary to share with your team or investors. The limitation isn't intelligence. It's that every run is a fresh session with no memory of last week, no live connection to your CRM, and no way to push updates back into your tools.

Skip the manual work → See the Starch version

Sales & CRM3 AI tools7 steps6 friction points

AI walkthrough

How to do it with AI today

A practical walkthrough using ChatGPT, Claude, and other off-the-shelf LLMs — what they're good at, what you'll have to do by hand.

Tools that work for this

ChatGPTClaudeGemini

Step-by-step

1 Export your current deal list from your CRM — HubSpot, Salesforce, Pipedrive, or a spreadsheet — as a CSV or copy the table. Include deal name, stage, expected close date, deal value, owner, and last activity date.

2 Open ChatGPT or Claude and paste the deal data with a framing prompt that tells the model what a 'healthy' deal looks like in your pipeline — your stage definitions, your typical sales cycle, and what counts as 'stalled' for your business.

3 Ask the model to categorize each deal: on track, at risk, stalled, or needs immediate action. Have it explain its reasoning per deal in one sentence so you can sanity-check before you act.

4 Run a second prompt asking for a pipeline summary you can share: total pipeline value, weighted forecast, deals closing this week, and top three risks. Specify the format — bullet list, table, or paragraph — depending on where you're pasting it.

5 Paste in any relevant email threads or call notes for your highest-value deals and ask the model to identify the key blocker or next step from that context. Claude handles long documents well here.

6 Use the model to draft follow-up emails or Slack messages for the deals flagged as stalled — give it the deal context and the last touchpoint, and ask for a short, direct message to re-engage the contact.

7 Copy the model's output into your CRM notes, a shared doc, or a Notion page manually. Nothing flows back automatically — every update is a copy-paste job.

Prompts you can copy

Here is my pipeline export as of today [paste table]. My sales cycle is 30 days. Any deal with no activity in 14+ days is stalled. Categorize each deal as on-track, at-risk, or stalled and give one sentence of reasoning per deal.

Summarize this pipeline in 150 words for a weekly investor update: total pipeline value, weighted forecast at 60% close rate, number of deals by stage, and the top two risks to hitting this month's number.

Here are the last three email threads with [Company Name] [paste emails]. What is the actual blocker right now, and what would you recommend I say in a follow-up this week to move this forward?

Draft a short re-engagement email to [contact name] at [company]. Context: we demoed two weeks ago, they went quiet, deal is $18k ARR. Tone: direct, not pushy. Under 100 words.

Based on this deal list, which three deals should I prioritize my time on this week if my goal is to close the most revenue before end of month? Explain your reasoning.

Reality check

Where this gets hard

The walkthrough above works — until your numbers change, the LLM hallucinates, or you have to re-paste everything next month.

No live CRM connection — you start every weekly review by manually exporting a CSV or copying a table, so the data is already slightly stale before you paste it.

The model has no memory of last week's review. You can't ask 'what changed since last Monday?' — there's no baseline to compare against unless you paste last week's export too.

Output format drifts run to run. The deal categorization table you got last Tuesday may come back as a bulleted list this Tuesday, requiring you to reformat before you can share it.

Nothing writes back. Every action item, updated stage, or follow-up note the model generates has to be manually entered into your CRM. The review creates work rather than completing it.

Context limits bite on large pipelines. Paste 80 deals with email history and you'll hit truncation, forcing you to split the analysis into chunks and stitch the summary together yourself.

There's no scheduling. The weekly review happens when you remember to run it — there's no system that pulls data and surfaces the analysis on Monday morning without you initiating it.

Tired of the friction?

Starch runs the whole workflow on live data — no copy-paste, no hallucinated numbers, no re-prompting next month.

See the Starch version →

Starch alternative

The same workflow on Starch

Starch is an agentic operating system — it builds and runs the software your pipeline review depends on, continuously, against your live CRM and email data. Instead of re-running prompts each week, you describe what you want once, and Starch builds a persistent app that does it.

Starch syncs your HubSpot deals, contacts, and activity data on a schedule — so when your pipeline review runs Monday morning, it's working from real current data, not a Friday export you copied by hand.

Connect Gmail or Outlook once, and Starch reads your email threads alongside deal records. Ask 'which deals have had no response in 14 days?' and get an answer that reflects actual inbox activity, not just CRM log entries.

Start from the Sales Agent CRM app in the Starch App Store — it connects HubSpot and Apollo and gives you a pipeline surface built for this exact review loop. Customize it by describing your stage definitions and what 'at risk' means for your sales cycle.

Describe the weekly report format you want in plain English — 'every Monday, summarize my pipeline by stage, flag stalled deals, and Slack me the three highest-priority actions for the week' — and Starch builds and schedules that automation.

Pipeline notes, deal stage updates, and follow-up drafts can flow back into your CRM through the same connected system — not a separate copy-paste step after you've finished reading the output.

Build a shared pipeline dashboard your whole team can see, pulling live from HubSpot, without a BI tool or an analyst to maintain it — describe what you want displayed, and Starch assembles it.

Get closed-beta access →

Toolkit