By Gerard Pietrykiewicz and Achim Klor
Achim is a fractional CMO who helps B2B GTM teams with brand-building and AI adoption. Gerard is a seasoned project manager and executive coach helping teams deliver software that actually works.
During a recent (and timely) Lenny’s Podcast episode, Jason Lemkin, cofounder of SaaStr, dropped a few jaws and rolled some eyes (just read the comments) with this:
He used to run go-to-market with about 10 people. Now it’s 1.2 humans and 20 agents. He says the business runs about the same.
That’s a spicy headline.
It’s also the least useful part of the story.
If you’re trying to adopt AI in GTM without breaking trust, below are field notes worth considering.
Jason’s most useful advice is not to “immediately replace your GTM team.”
It’s the Incognito Test:
You will probably find something broken. Like maybe support takes too long. Maybe the contact “disappears” in the CRM ether. Maybe your SDRs take three days to respond (or never).
Pick the thing that makes you the most angry. That’s your first agent use case.
One workflow. One owner. One month of training. Daily QA.
Then decide if you need a second, a third, and so on.
This is a pattern Gerard and I keep seeing too.
The real question isn’t “What’s our AI strategy?” It’s “What’s the smallest thing I can do right now that actually helps?”
Source: Lenny’s Podcast, 01:25:29
Two salespeople quit on-site during SaaStr Annual. No, not after the event… during.
Jason Lemkin had been through this cycle before. Build a team, train them, watch them leave, rinse and repeat. This time, he made a different call:
“We’re done hiring humans in sales. We’re going all-in on agents.”
Fast-forward six months: SaaStr now runs on 1.2 humans and 20 AI agents. Same revenue as the old 10-person team.
Source: Lenny’s Podcast, 0:11:52
Jason said this about SaaStr’s “Chief AI Officer”, Amelia* (a real person):
“She spends about 20%* of her time managing and orchestrating the agents.”
Every day, agents write emails, qualify leads, set meetings. Every day, Amelia checks quality, fixes mistakes, tunes prompts.
This isn’t “set it and forget it.” It’s more like running a production line. A human checks it, fixes it, and tunes the system so it does not drift.
Agents work evenings. Weekends. Christmas.
But if nobody’s watching, the system decays… fast.
And that’s the part vendors don’t put in their demos.
* Amelia is the 0.2 human who spends 20% managing agents. Jason is the 1.0.
Jason’s breakthrough came from a generic support agent. It wasn’t built for sales. It wasn’t trained on sponsorships. But it closed a $70K deal on its own because it “knew” the product and responded instantly at 11 PM on a Saturday.
That’s why this conversation matters.
Not because an agent can write emails. But because one can close revenue if you give it enough context and keep it on rails.
Source: Lenny’s Podcast, 0:07:53
Jason’s experience lines up with what Gerard and I have also seen in the field.
Start with a real workflow, not an “AI strategy”
Speed does not fix the wrong thing
Adoption is a leadership job
Whenever new tech comes along, almost everybody immediately wants the potential upside.
Almost nobody wants the governance.
It reminds me of the Dotcom boom. Everyone wanted a website. No one wanted to manage the content ecosystem (that’s still true today).
Agentic AI, like what Jason implemented at SaaStr, is no different.
Customer-facing GTM output carries real risk for your brand reputation. If your AI agents hallucinate, spam, or misread context, you don’t just lose a meeting. You lose trust.
And that’s the simple truth we keep coming back to:
The constraint is not capability. It’s trust. And trust is earned, not bought (or a bot).
In our AI adoption challenges article, we said tool demos hide the messy middle: setup, access, QA, and the operational reality of keeping this stuff reliable.
Jason’s story doesn’t remove that mess. It confirms it. He just decided the mess was worth automating.
Watch the full conversation: Lenny's Podcast with Jason Lemkin
If you like this co-authored content, here are some more ways we can help:
Cheers!
This article is AC-A and published on LinkedIn. Join the conversation!
