Free prompt templates

Prompts for revenue teams that want the evidence, not the story.

Copy any of these into Claude, ChatGPT or the assistant you already use. Each one checks a deal, a call or a document against what was actually said or written, and asks for the quote behind every claim. No account needed.

Check your company’s policy before pasting customer calls or CRM data into an AI tool.

Deals and forecast

Check your forecast against what buyers said

Use it when
Before a forecast call, when a close date rests on someone’s optimism rather than the buyer’s words.
You need
Your forecast or pipeline export, and the latest call transcript for each deal you want to check.
You are checking a sales forecast against the evidence.

For each deal below I give you the CRM record (close date, stage, amount, forecast category, next step) and the transcript of the most recent call.

For every deal:
1. Quote anything the buyer said about timing, budget, sign-off or competing priorities, with the speaker's name.
2. Say whether those quotes support the CRM close date and forecast category. Answer "supported", "contradicted" or "not discussed".
3. If contradicted, say which field is wrong and what the call suggests instead.

Do not guess. If the transcript does not mention something, say "not discussed". Finish with a table: deal, CRM close date, what the buyer said, verdict.

Deals:
[paste each CRM record and its transcript here]

With ArciveArcive checks this every morning: when a buyer says something the CRM close date does not reflect, the deal owner gets it in a short brief with the quote and the moment in the call.

Find where a CRM record and the last call disagree

Use it when
When you are about to rely on a deal record and are not sure it was updated after the last call.
You need
One CRM deal record and the transcript of the latest call on that deal.
Compare this CRM deal record with this call transcript.

List every place where the call contradicts or goes beyond the record: close date, next step, amount, decision makers, blockers, competitors, anything promised by either side.

For each one give:
- the CRM field and its current value
- the exact quote from the call, with the speaker
- what the field should probably say now

If the call and the record agree, say so. Do not invent fields that are not in the record.

CRM record:
[paste]

Call transcript:
[paste]

With ArciveArcive matches each call to its deal and checks the record against what was said, so the owner hears about a next step agreed on a call that the CRM has not caught up with.

See what changed on a deal since the last review

Use it when
Before a pipeline review, to spend the time on what moved rather than re-reading everything.
You need
The deal record as it was at the last review, the record now, and any calls since.
Here is a deal record from our last review, the same record today, and the transcripts of calls since then.

Tell me, in this order:
1. Which fields changed (close date, stage, amount, forecast, next step), from what to what.
2. What was said on the calls that matters for the deal, with quotes and speakers.
3. Anything the calls say that the current record does not reflect.
4. One question I should ask the deal owner in the review.

Keep it under 200 words. Do not repeat anything that did not change.

Record at last review:
[paste]

Record today:
[paste]

Calls since:
[paste]

With ArciveArcive’s daily deal brief tells the owner when the close date, stage, amount or forecast moves, alongside what the calls said.

Work out which CRM fields you can actually trust

Use it when
When reports look wrong and you suspect the data under them, not the maths.
You need
An export of open deals with their fields and last-modified dates, plus a handful of recent call transcripts.
I am auditing how far our CRM can be trusted.

From the deal export:
1. For each field (next step, close date, amount, stage, forecast category), give the share of open deals where it is empty, and where it was last changed more than 30 days ago.
2. List deals whose close date is already in the past.
3. List likely duplicates: same company, similar names, overlapping amounts.

From the call transcripts:
4. For each deal that has a transcript, say whether the call supports the record's next step and close date.

Finish with a short list of the three fields we should stop relying on until they are cleaned up, and why.

Deal export:
[paste]

Transcripts:
[paste]

With ArciveIn its first week Arcive reports duplicate accounts and people, sources that disagree, and where calls and the CRM part ways.

Calls

Pull the promises out of a call before they are forgotten

Use it when
Straight after a customer call, or when picking up an account someone else ran.
You need
A call transcript with speaker names.
Read this call transcript and list every commitment made by either side.

For each commitment give:
- who made it and who it was made to
- the exact quote
- what was promised
- the deadline, as stated ("by Friday", "next quarter"), and as a date if the call date makes it clear
- whether it is ours or the customer's

Then list anything that sounds like a commitment but has no owner or no date, so someone can follow up.

Call date: [date]
Transcript:
[paste]

With ArciveArcive keeps each promise made on a call with its quote and due date, and reminds the deal owner when one of ours is due or overdue.

Build a stakeholder map from your calls

Use it when
When a deal has had several calls and you want to know who actually matters on the buyer’s side.
You need
The transcripts of every call on the deal, oldest first, with dates.
From these call transcripts, build a map of the people on the buyer's side.

For each person give:
- name and title, if stated
- which calls they joined
- what they care about, in their own words (quote it)
- their apparent role in the decision: decision maker, influencer, user, blocker, or unclear

Then tell me:
- who appeared most recently for the first time
- who used to join and has stopped
- whose sign-off the calls suggest we still need

Only use what is in the transcripts.

Transcripts:
[paste, oldest first]

With ArciveArcive matches the same person across calls and CRM into one record, and tells the deal owner when someone new from the buyer’s side joins a call.

Write a handover note when an account changes owner

Use it when
When someone leaves, changes patch, or passes an account to customer success.
You need
The account and deal records, recent call transcripts, and any open email threads.
Write a handover note for the new owner of this account.

Use only the material below. Structure it as:
1. Where this stands now, in one paragraph.
2. People on their side, and what each cares about (quote where you can).
3. What we have promised and by when.
4. Open risks and blockers, with the evidence for each.
5. What the CRM says that the calls do not support.
6. The first three things the new owner should do.

Mark anything you are unsure about as "unconfirmed". Do not fill gaps with assumptions.

Material:
[paste]

With ArciveWith Arcive, the new owner can ask about the account and get the answer with the call, record or document it came from, without rebuilding it from scratch.

Documents and answers

Find where your sales documents contradict each other

Use it when
When a pricing deck, a policy page and an email template might each say something different.
You need
The documents themselves: pricing, discount rules, product sheets, playbook pages.
Read these documents and find every place where two of them disagree.

Look especially at prices, discounts, approval rules, contract terms, what we can promise and product limits.

For each disagreement give:
- the two statements, quoted, with the document each comes from
- which one is more recent, if the documents say
- what a customer would hear if two people used different documents

Do not flag differences in wording that mean the same thing.

Documents:
[paste each with its title and date]

With ArciveArcive flags where calls, CRM records and documents disagree, and warns inside any answer that rests on sources that disagree.

Turn the questions your team keeps asking into one written answer

Use it when
When the same question lands on a manager every week.
You need
A list of questions people have asked, copied from wherever they were asked, and the documents you already have.
Here are questions people on my team have asked recently, and the documents we already have.

1. Group the questions that are really the same question. Name each group in plain words.
2. For each group, say whether our documents answer it. Quote the answer and its document if they do.
3. For each group our documents do not answer, draft the short answer someone should write down, and list what a manager would need to confirm before it is published.

Order the groups by how often they were asked.

Questions:
[paste]

Documents:
[paste]

With ArciveArcive notices the questions your team keeps asking that no source answers, and turns them into a content request so the answer is written down once.

Tools and AI

Decide what your AI assistant should be able to read

Use it when
When you want Claude, ChatGPT or another assistant to work across your company’s tools, not just one chat.
You need
A list of the tools your revenue team uses, and what each one is used for.
I want an AI assistant to help my revenue team using our internal tools. Help me plan what it should be able to read.

For each tool I list:
1. What it is the source of truth for (for example: calls for what the buyer said, the CRM for deal stage and amount, the wiki for pricing rules).
2. Where it overlaps with another tool, and which one should win when they disagree.
3. Who owns keeping it correct.
4. Who in the team should be able to see what is in it, and anything that must never reach the assistant.

Then give me:
- the three connections that would answer most of our questions
- the questions that would still go unanswered because nothing holds the answer
- the risks of an assistant reading two tools that disagree without knowing which is current

Our tools:
[list each tool and what it is used for]

With ArciveArcive connects Gong, HubSpot, Notion, Google Drive, Confluence and Google Calendar, shows where they disagree, and lets Claude and other AI tools read what it knows over MCP, with the permissions of the person asking.

Make an AI assistant show where each answer came from

Use it when
When people stop trusting assistant answers because they cannot tell what they rest on.
You need
An assistant with your documents or tools attached.
From now on, answer my questions about our customers, deals, pricing and policies only from the sources I have given you.

For every answer:
- cite the source for each claim: document title, record, or call and speaker
- say how recent each source is, if you can tell
- if two sources disagree, show both and say which looks more current and why
- if no source answers the question, say "not in the sources" instead of answering from general knowledge

Keep answers short. Put the citations at the end of each sentence they support.

With ArciveEvery Arcive answer carries its sources and how fresh each one is, and warns when the sources behind it disagree.

Doing this by hand every week?

Arcive keeps it checked across your calls, CRM and documents, and tells the person who owns the deal when something no longer adds up.

Request a demo