Saster

Jason Lemkin started Saster almost by accident. In 2012, after selling his startup to Adobe, he began writing a blog documenting the mistakes he'd made as a founder. This resonated with the community. He organized meetups—before it became trendy—that grew into Saster Annual, now attracting 10,000 people yearly. Over time, he invested nearly $200 million back into founders from the community. What began as a passion project evolved into an eight-figure revenue business, with sponsorships averaging $70-80K and ticket sales ranging from a few hundred to $2,000 for VCs.

For years, Saster ran a traditional go-to-market machine: 2-3 SDRs and up to 5 AEs managing outbound email campaigns, inbound qualification, and sponsorship sales across a database of 400,000+ prospects. This was expensive, high-touch work. Then in May 2024, something changed. Two of Saster's highest-paid sales reps quit on-site at the annual event. Frustrated by repeating this cycle for the eighth time, Jason made a radical decision: no more hiring junior SDRs.

Jason had been experimenting with Delphi, an AI digital clone tool, to create "digital Jason" for customer support and general inquiries. One day, this untrained agent—not built for sales—closed a $70K sponsorship deal on its own. That moment was the catalyst. Jason realized if a general-purpose agent could close deals, purpose-built sales agents could do even more. He partnered with Artisan (a YC-backed startup that offered to help, unlike vendors demanding $100K upfront) to build an outbound agent. It sent 60,000 emails with high response rates to lapsed prospects. Then he deployed Qualified for inbound qualification, which immediately worked: a prospect at 11pm Saturday night wanted to sponsor, and the AI closed it.

The biggest lesson: don't build it yourself unless you're Vercel. Saster didn't. Instead, Jason chose vendors based on who would help most, not just features. This required picking a "forward deployed engineer" (solution architect) willing to roll up their sleeves. Training the agents was critical—uploading company docs, answering questions, correcting mistakes daily. After 30 days of 1-2 hours of daily iteration, the agents became genuinely good. Jason also learned that email quality depends entirely on training: copy your best salesperson's template, let the AI iterate and A-B test variants, and personalize with data from Salesforce or website visitor tracking. People don't care if it's AI—they care if it adds value and gets instant responses.

Today, Saster's office is quiet. Ten desks labeled with agent names (Replete, Quality, Artisan, AgentForce) sit empty of humans. Amelia, the Chief AI Officer, manages 20 agents in her corner office. Jason spends 10-15 hours weekly reviewing agent outputs. The business does similarly to when it had 10 humans, but with 1.2 humans instead. Agents work nights, weekends, and Christmas—no vacation. Jason sees the future clearly: SDRs who send emails will be 90% displaced next year. BDRs who qualify inbound leads should be extinct. AEs will see 40-50% of their jobs at risk. But the best humans—those managing and orchestrating agents—will become superpowers. His advice to anyone worried: pick one tool, train it yourself for 30 days, and become hyper-employable. The vendors that helped Saster most will help you. The window for deploying agents closes as enterprises get exhausted from implementing too many—but for now, everyone in market and adoption is the name of the game.