Via Lenny’s Podcast
On shifting to an agentic sales team: “The net productivity is about the same. It’s not better. It’s not worse. But it’s so much more efficient and it scales because software scales. […]
I think it’s important [to note] that it takes time to train these agents. They don’t work out of the box. But when you dial them in, when you take your best person or your best script and you train an [AI] agent with your best person and best script, that agent can start to become a version of your best salesperson, your best person.
— We replaced our sales team with 20 AI agents—here’s what happened next | Jason Lemkin (SaaStr), at 9:17
Jason Lemkin of SaaStr: the key to getting effective AI Sales Agents is training them on your absolute best performers and most refined sales scripts.
“When you take your best person or your best script and you train an agent with your best person and best script, that agent can start to become a version of your best salesperson.”
Training methodology:
- Document top performer calls and email sequences
- Extract the specific language, objection handling, and qualification questions that work
- Program agents with this optimized approach rather than indexing on average team behavior
- Iterate based on results, continuously raising the performance bar
This means AI can actually democratize best practices—every prospect gets your “A-game” instead of whichever random sales rep picks up the lead.
However, SaaStr doesn’t use one general-purpose agent—they deploy different specialized agents for different sales scenarios. (at 11:31)
SaaStr’s agent architecture includes:
- High-end sponsorship agents: Handle $70-80K average deals with longer sales cycles
- Ticket sales agents: Manage high-volume, lower-price transactions ($200-$2,000)
- Lapsed customer agents: Specialize in reactivation and win-back campaigns
- Qualification agents: Handle inbound “contact me” leads
- Outbound agents: Run cold outreach campaigns
“We have lapsed high-end and low-end agents. And they have different workflows and we actually use different vendors for now.”
The specialization mirrors how you’d structure a human team—different roles for different motions—but with the advantage that agents can be precisely configured for narrow use cases.
Leave a comment