Three foundation models graded 120 B2B APIs and agreed: marketing automation and sales engagement platforms have no place in agent workflows. When AI can craft and send personalised emails natively, the templating layer that defined these categories becomes redundant. Here is what that means for your stack.
Jason Lemkin opened the books on SaaStr's AI agent stack. Three humans, 20+ agents, 614 meetings booked, and $500K recovered from leads no rep would touch. Real costs, real output, and the one place you should not point an agent.
Jason Eubanks put his old GTM stack on a slide: 22 products, $3 million in annual fees, 11 ops team members to keep it running. Reps spent 24-30% of their time actually selling. The rest was context switching across 10-12 tools per day. Now he is building Aurasell to replace the whole thing with one AI-native CRM.
Jason Lemkin's SaaStr runs 20+ AI agents with 3 full-time humans, same revenue as when they had 20 employees in 2020. Their custom-built AI VP of Marketing, called 10K, plans campaigns, manages pipeline, and assigns daily tasks. They spent over $500k on AI infrastructure because every AI marketing tool they demoed only wrote content, not strategy.
The B2B media company rebuilt its entire go-to-market stack around AI agents in six months. Two humans and a dog now manage sponsor sales, event production, and customer success through specialized agents that book meetings, revive dead leads, and push daily marketing ideas. Owner.com and Klaviyo shared similar AI agent buildouts at SaaStr AI Annual 2026.
Jason Lemkin's SaaStr AI team replaced most of their go-to-market function with AI agents. The stack handles outbound, inbound, and RevOps across 2.25M sessions. Here is what worked, what broke, and the actual numbers behind each agent.
Meta fixed a security flaw that let attackers manipulate its AI support bot to take over Instagram accounts, including high-profile business profiles. The exploit worked by social-engineering the chatbot into sending password-reset codes to attacker-controlled emails. Meta did not disclose how many accounts were affected.
Jason Lemkin's SaaStr is running four specialized AI SDR platforms instead of one all-in-one tool. Three Artisan instances for outbound (40,000+ messages sent), Qualified for inbound ($1M+ closed), Agentforce for reactivation (200,000 messages, 72% open rates), and Monaco for net new logos. The take: specialized tools beat generalist platforms once you hit scale, but most companies should start with one.
Anthropic's Head of Industries walked through their GTM stack at SaaStr AI. No custom platform. They use Clay, LeanData, Salesforce, Gong, Ironclad, and Slack. The difference: Claude does the data entry, qualification routing, and call prep. 54% of enterprise logos came through AI-qualified self-serve in 2026.
SaaStr's Qualified-powered agent handled 2.2M sessions and booked 614 qualified meetings at its 10,000-person conference. Founder Jason Lemkin says B leads, not A leads, are where AI ROI actually shows up. Here is what the conversion data looks like and why B2B teams should care.
CBA is testing an AI agent that handles business loan applications end-to-end, including income verification. The tool lives in the mobile app and uses customer data to answer questions and progress applications without human involvement. This is not a chatbot upgrade, it is agentic AI that takes action.
SaaStr is running two AI agents that handle marketing ops, sponsor management, and personalized outreach for $257 combined monthly cost. The work previously required multiple full-time staff. LLM costs dropped 95% in 12 months, making agent economics viable for lean GTM teams.