ye myat
book intro →

AI Agents for Lead Warming

Re-engage cold leads with personalized outreach based on recent company news and your last conversation. No more 'just checking in' emails.

01the problem

Your CRM has dozens of leads that went cold. You know you should follow up, but writing a personalized message for each one takes research: what did we last talk about? What's new at their company? You end up sending nothing or worse, a generic 'just checking in' that gets ignored.

02what the agent does

Every Monday at 10 AM, the agent finds leads inactive for 45+ days in your CRM. For each one, it pulls your last email conversation, recent Slack mentions, and runs a web search for company news (funding, leadership changes, product launches). Then it drafts a personalized warming message that references something specific, not a generic follow-up.

Example output

agent output

Jake Torres — Acme Corp

  • Last activity: Jan 22
  • Deal stage: Qualified
  • Last conversation summary: Discussed Q1 budget for marketing automation
  • Recent news: Acme announced Series B ($12M) on Feb 28
  • Draft message:

    Hi Jake, saw the Series B announcement. Congrats. When we spoke in January you mentioned Q1 budget was tight for the automation project. With the new funding, does it make sense to revisit? Happy to do a 15-minute call to see if the timing works now.

Try it yourself

Copy these prompts into Claude Cowork. Connect your tools. The agent handles the rest.

01
Weekly lead warming
Runs every Monday. Surfaces cold leads with personalized re-engagement messages.
Run every Monday at 10:00 AM.

1. Pull contacts from HubSpot where last activity > 45 days ago, deal not Closed Won or Closed Lost
2. Search Gmail for recent emails with each contact
3. Search Slack for mentions of contact or company name
4. Web search for recent news about each contact's company
5. Draft a warming message per lead: reference something specific, no generic openers, end with one low-friction ask
6. Save report to projects/leads/YYYY-MM-DD-warming.md
7. Email to me for review. Do not send to leads.

FAQ

Common questions about setting up this agent.

How does the agent find company news?+
It runs a web search for each lead's company, looking for press coverage, funding announcements, leadership changes, or product launches from the last 30 days. If nothing is found, it falls back to referencing your last conversation instead. Either way, the message is specific to that person.
What CRMs does this work with?+
Claude Cowork connects to HubSpot, Salesforce, and other major CRMs. If you use a simpler setup (like a spreadsheet or Notion database), the agent can read from that too. The key is having lead contact info and last activity dates somewhere accessible.
Won't leads know this is AI-generated?+
Not if the message references something real. A generic 'just checking in' feels automated because it is. A message that says 'saw your Series B announcement, want to revisit the automation project we discussed?' feels personal because it's built from actual context. The agent does the research. You review the output.

More for startup founders

Daily Planning

Start every morning with a prioritized work plan built from your calendar, Slack, email, and project tools. No manual assembly required.

view

Invoice Chasing

Automatically detect overdue invoices, draft follow-up emails, and escalate the ones that need attention. No more awkward manual chasing.

view

Weekly Metrics Reports

Pull MRR, churn, burn rate, and runway from Stripe and your accounting tool every Monday. Formatted, compared to last week, and in your inbox before standup.

view

Want this running for your team?

I set up AI agents like this for startup founders. 30 minutes, no strings.

book intro →