How to Use AI with HubSpot for Sales

One time, about 20 minutes

Help me write a one page HubSpot field guide for my AI assistant. Ask me these questions one at a time, then write the guide under the headings below:
1. What are our deal pipeline stages, and what has to be true for a deal to move to each one?
2. Which deal properties matter most, and what does a good value look like for each?
3. Which contact and company properties do I use to decide if a lead is a good fit?
4. What rules do managers check in pipeline reviews?
5. What must AI never change or assume?

Headings: Stages and exit criteria · Key deal properties · Fit properties for leads · Review rules · Never do

Weekly, before your pipeline review

Using my HubSpot field guide and the deal list below (name, stage, amount, close date, next step, next step date, last activity date), find:
1. Deals with no next step or a next step date in the past
2. Close dates in the past or moved more than twice
3. Deals whose stage doesn't match the exit criteria in my field guide, based on the notes
4. Deals with no activity in 14 days
For each, give the deal name, the problem, the property it's based on and the one question I should answer to fix it. Don't suggest values I haven't given you. Write MISSING where data is missing.

<deals>
{paste your export}
</deals>

Right after each call

Here's the transcript (or my notes) from a call with {company}. Using my field guide, draft:
1. A deal note: 3 sentence summary, the buyer's priorities in their words, risks.
2. Property updates: only properties the call gives evidence for, each with the quote that supports it.
3. Tasks: every next step, who owns it (us or them) and the date said on the call.
4. A recap email under 120 words.
If something wasn't discussed, write NOT DISCUSSED. Mark any date or owner that isn't clear as UNCLEAR.

<transcript>
{paste transcript or notes}
</transcript>

When new leads come in

Using the fit properties in my field guide, sort these leads into three groups: top priority, worth a lighter touch and not a fit right now. For each lead, give the group, the two properties that decided it and anything missing that would change the answer.
Don't research outside what I've pasted. Don't mark anyone "not a fit" because of missing data; put them in a fourth group, NEEDS INFO.

<leads>
{paste lead export: company, title, industry, size, source, notes}
</leads>

The day before your review

Act as my sales manager preparing for our weekly pipeline review. Using my field guide's review rules and the deals below, list the 5 questions you'd ask me first, each tied to a specific deal and property. Then, for each question, draft a two sentence answer using only what's in the data, and flag where the data doesn't support a confident answer.

<deals>
{paste your open deals with notes}
</deals>

For deals with a buying group

Here are the contacts associated with the {company} deal and our activity with each (emails, meetings, calls, last date). Build a coverage table: name, role, what they care about based on our notes, last contact date and strength (strong, weak or none).
Then list the roles a deal like this usually needs that are missing from the list. Don't invent people; list missing roles, not names.