Jason Lemkin, often referred to as the Godfather of SaaS, believes the moment has arrived to push AI’s capabilities in the workplace to new extremes. Lemkin, founder of SaaStr—the world’s largest community of business-to-business founders—recently shared on Lenny's Podcast that this shift means he will cease hiring humans for his sales team entirely. Instead, SaaStr is fully embracing AI agents, which are virtual assistants capable of autonomously completing tasks. These agents analyze problems, devise strategies, and take independent actions without needing user prompts. Currently, Lemkin revealed that SaaStr employs 20 AI agents performing tasks previously managed by a team of 10 sales development representatives and account executives. This transition from a fully human sales force to an agent-driven team happened quickly. Back in May, SaaStr had only a single AI agent deployed, handling various digital functions, Lemkin noted. That same month, however, during the SaaStr Annual—an event attended by over 10, 000 founders, executives, and venture capitalists—two of its highest-paid salespeople suddenly resigned. Following this, Lemkin approached the company’s chief AI officer, stating, “We’re done hiring humans in sales.

We’re going to push the boundaries with agents. ” His reasoning was straightforward: it wasn’t worthwhile to recruit another junior sales rep at a $150, 000 annual salary who would likely leave eventually, when a reliable AI agent could fill the role instead. Amelia Lerutte, SaaStr’s chief AI officer, shared with Business Insider via email that by June, the company started significantly increasing the number of agents in operation. “At that time, we had only one non-core agent named Delphi, but didn’t expand from 2 to over 20 until early June, ” she explained. “The decision to redirect some (though not all) human headcount budget to agents was deliberate following those departures. ” At SaaStr’s office, the ten desks formerly occupied by human members of the go-to-market team now display the names of agents such as “Quali for qualified, ” “Arty for artisan, ” and “Repli for Replit, ” Lemkin said. He added that SaaStr trains its AI agents using the expertise of its top human performers. “By training an agent with your best salesperson and their best script, that agent can evolve into a version of your strongest rep, ” he commented. This approach mirrors that of Vercel, the cloud platform for developers, which trained a sales agent by meticulously documenting all the steps of its star salesperson’s methods over six weeks, then creating an agent to replicate that workflow. While many companies are exploring AI agents, there remain risks, notably the dangers of data breaches and cybercrime. “AI agents, to function fully and access necessary applications, often require entry to the device’s operating system or OS-level access, ” warned Harry Farmer, senior researcher at the Ada Lovelace Institute, in a recent interview with Wired. Such extensive access introduces additional vulnerabilities for potential cyberattacks. Despite security concerns, Lemkin noted that agents’ overall productivity matches that of humans. Yet, agents offer greater efficiency and scalability, functioning similarly to software.