AI Sales Plays

Today’s plays › All plays › Follow-up

Play · Follow-up
For sellersFor sales leadersHubSpot Prospecting Agent (Breeze)

Log every edit to AI-drafted emails, and turn on auto-send only when the edit rate beats your bar

Picked and checked by Olga Pearson, SalesUpLevel

Get tomorrow’s play in your inbox. Free ↓

The resultas reported

Reported by HubSpot and Stax Payments: under 3% of AI-drafted emails were edited against a 20% target, so Stax turned on auto-send. Automated inbound lead outreach got a 65% action rate (a reply or a booked meeting), automated cold outbound a 6% action rate, and the BDR got back about 8 hours a week of research and writing.

Most teams argue about whether AI emails are good enough to send. Stax settled it with one number they tracked from day one.

Case study: Stax Payments (customer) and HubSpot (vendor). One BDR covered cold prospecting, inbound follow-up, event follow-up and re-engagement for partner sales. Stax launched the agent in October 2025 in draft-and-review mode. They started with one willing BDR and about 500 contacts enrolled by hand, and he logged in a custom CRM field whether he edited each draft before sending. The bar was 20%. The result was under 3% (Julia Olson's post says 2 to 3%). They built a separate play with its own instructions for each use case, with more urgency for hot inbound leads, and gave the agent account facts such as the prospect's current provider.

By Julia Olson and Ashley Forbes (Stax Payments), in a HubSpot case study, Senior Director of CRM Strategy & Operations (Julia Olson) and Product Manager (Ashley Forbes), Stax Payments · Case study (HubSpot customer story), plus Julia Olson's LinkedIn post ↗ · Oct 30, 2026 (Julia Olson's LinkedIn post; HubSpot page has no date)

Run it before your 10am

  1. Pick one willing BDR and a few hundred contacts, and keep the AI in draft-and-review mode.
  2. Add a CRM field: edited before sending, yes or no.
  3. Set your bar before you start (Stax used 20%).
  4. Write separate instructions for each use case: cold, hot inbound, event follow-up, re-engagement.
  5. Give the AI one or two account facts that matter, like the prospect's current provider.
  6. Turn on auto-send only for a use case that beats your bar, and keep checking a sample.
Copy-paste prompt · Try it tomorrow (our copy-paste prompt; the case study does not share Stax's instructions)
You write first emails for [company] to a [use case: hot inbound lead / event attendee / cold prospect].
Contact: [name, title, seniority]. Company: [company]. What they did: [form filled / session attended / nothing yet]. Current provider: [provider]. Past notes: [CRM notes].
Write under 100 words. Shorter for executives, more detail for hands-on roles.
Use one fact from their history with us. Do not reintroduce our company from scratch.
For hot inbound leads, offer two times to talk today or tomorrow.
End with one question.

If you lead a team

  1. Agree the bar with sales and legal before the test starts.
  2. Turn on auto-send one use case at a time, and keep a weekly sample check.

Watch out for

  • Vendor case study; HubSpot is promoting its agent, and Julia Olson's post ties it to HubSpot's Loop Marketing framework.
  • "Action rate" counts replies and booked meetings together. 65% is warm inbound leads; cold outbound was 6%.
  • One BDR and about 500 contacts to start. Your edit rate may differ.
  • What I'd have my team try first: one BDR, 100 inbound leads, drafts only, and the edited yes or no field.

Sources: HubSpot case study 'How Stax Payments Earned a 65% Action Rate by Using HubSpot's Prospecting Agent to Address Prospect Challenges' (no date on page; describes an October 2025 launch) and Julia Olson's LinkedIn post (Oct 30, 2026). Case study: customer Stax Payments, vendor HubSpot. Results are as reported; we did not measure them.

Run these next

Live cohort · SalesUpLevel

The AI-Native Seller

An 8-week live cohort, Wednesdays, starting Nov 11, 2026. Turn plays like these into how you and your team sell, every week.

Get in the room
Coming soon

Hiring AI-native sellers?

For sales leaders building a team that already works this way. More soon.