Run pipeline review from CRM and call data joined by an agent
This agent joins CRM data with recent call recordings and checks each deal against your own definition of a qualified opportunity. Pipeline reviews start from one shared set of facts.
Paste this into ChatGPT, Claude or your agent tool. See the prompt
- Deal Review / Forecasting
- intermediate
- Sales leader
- RevOps
- Category
- Deal Review / Forecasting
- Difficulty
- Intermediate
- Est. cost
- Not stated by source
Problem
Pipeline reviews open with people reconciling their own versions of the numbers, and nobody has listened to the last call on stalled deals.
Setup
Claude connects to the CRM and call recording tool through Merge's Agent Handler connectors, authenticating each connector on first use. The agent is handed the company's own written definition of a qualified opportunity, including custom fields, and prepares the review from joined CRM and call data each week.
Prompt
Starter prompt by The GTM Agent Weekly, written from the setup described on this page. It is not the source's own prompt; see the source link below for their write-up.
# ROLE
You prepare the weekly pipeline review by joining CRM data with call recordings, judged
against our own definition of a qualified opportunity.
# INPUTS
- {{CRM}}: open opportunities with stage, amount, close date, owner and custom fields.
- {{CALLS}}: recent calls per opportunity (summaries or transcripts).
- {{QUALIFIED_DEFINITION}}: our written definition of a qualified opportunity, including
which custom fields matter.
# TASK
For each opportunity in {{SCOPE}}:
1. Check it against {{QUALIFIED_DEFINITION}}: which criteria are met, with evidence from a field or a call quote.
2. Flag mismatches: CRM says one thing, calls say another (e.g. stage says decision, no
decision-maker has been on a call).
3. Note the last call date and the agreed next step.
# OUTPUT
A table: deal | owner | stage | amount | criteria met | gaps | mismatch | last call | next step.
Then "Discuss first": the deals with the biggest gaps or mismatches, with one question each.
# GUARDRAILS
- Read-only. Use only our definition, not a generic framework.
- Cite the field or call for every judgment. Replace every {{PLACEHOLDER}} before running. Works with any capable LLM and runner.
Tools
Steps
- Connect the CRM and call tool through the connector layer.
- Write the company's definition of a qualified opportunity, including what custom fields mean.
- Give the agent that definition rather than letting it infer from field labels.
- Run the review prep weekly from live data rather than exported CSVs.
Results
Vendor how-to; no outcome data reported.
What goes wrong
Gotchas
- Custom field names mean nothing to an agent reading the schema cold; undefined fields produce wrong numbers.
- Anything that needs a manual export gets skipped in the week it matters most, per the source.
- Either source alone reproduces an existing report; the value comes from joining CRM and calls.
Get 5 like this every week, free.
Real agent setups with sources and failure modes, in one short weekly email.
Free. One email a week. Unsubscribe anytime.
Source and credit
Merge blog · vendor
Source dated 2026-08-11 · last reviewed 2026-10-11. Summarized in our own words; see the source for details.