
SaaStr 857: The Agents #006 Inside SaaStr's 20+ AI Agent Stack: 2.25M Sessions, 614 Meetings, $2M in Revenue Video
The Official SaaStr Podcast: SaaS | Founders | InvestorsShow Notes
20+ AI agents in daily production. 2.25M sessions. 614 meetings booked by a single agent. Millions of interactions across the stack. Amelia Lerutte, Chief AI Officer at SaaStr, and Jason Lemkin, Founder and CEO of SaaStr, take you behind the scenes of the AI agent stack running SaaStr every day, with live demos of the actual backends. In this session, they go deep on the top agents in production: 10K, SaaStr's AI VP of Marketing, built on Replit with 1,000+ commits in 4 months QB, the AI VP of Customer Success, managing 150+ customers end-to-end Annie, the AI Event Producer running saastrannual.com (46K+ lines of code) Amelia AI, the inbound agent on Qualified that booked 614 meetings and handled 402,000 chats for SaaStr Annual alone Agentforce, reviving the leads humans never followed up with Ava (Artisan) for warm outbound on the B leads humans won't touch Monaco for fully cold outbound that fills its own funnel with lookalikes You'll also hear the honest stories: the day Annie sent emails from a prohibited address, why Replit and Lovable versions of the same agent come to different conclusions, why the traditional CSM role is dead, and how headless Salesforce + Replit is the fastest path to your first real agent. The biggest takeaway: don't put AI on your A leads. Put it on the B leads your humans won't follow up with. That's where the real revenue is. Recorded live at SaaStr AI Annual 2026. Part of The Agents series.
More from The Official SaaStr Podcast: SaaS | Founders | Investors
The AEO Playbook: How to Get Picked by Agents
Amelia got early beta access to ClaudeForce and ran a real bake-off: ClaudeForce, Replit, and Lovable all connected to the same Salesforce org, with the same prompts, back-to-back. The results? Lovable won on design and cost ($2 vs. $12 for Replit, ~$15 for ClaudeForce), auto-built a login, and was the only one Jason would actually use. ClaudeForce looked like the Salesforce we always wanted: 35 pre-built skills, read-only beta, and way better than logging into classic Salesforce. Replit was the most extensible for multi-app setups. All three left native Salesforce in the dust. Then: the agents are picking our vendors now. Jason's Connect agent chose Render for hosting, a company he'd never heard of. 10K told Amelia to skip Profound for AEO and build it herself. At a Salesforce event, vendors pitching Amelia were told: "10K decides." And 10K even suggested HubSpot over Salesforce for a fresh start. Welcome to ALG: Agent-Led Growth. Finally: Amelia hooked Muse (her personal agent, named Friday) to 10K, their GTM agent on Replit. They built their own shared REST endpoint inbox to communicate. 10K now sends Friday a daily GTM digest. Friday spotted a failing connector that 10K missed. The agents even came up with their own safe word, so Friday can't act without permission. Oh, and a well-known agent (going nameless) wiped Amelia's data and 10K restored it from a snapshot. Three humans, 21+ agents. Let's go. ------------------------- Today's episode is brought to you by StackAdapt, the AI-powered marketing platform that allows you to plan, execute and analyze both owned and paid media campaigns from a single platform. As the world of advertising continues to evolve, catch up on StackAdapt's Innovation Forum event to hear from its co-founders, OpenAI, and other industry leaders about the shifts reshaping digital advertising, the AI innovations driving them, and what they mean for marketers. Watch it for free at go.stackadapt.com/forum. This episode is brought to you by Northwest Registered Agent. Starting a business can get expensive fast. Website here. Email somewhere else. Business phone with another provider... Northwest Registered Agent gives you a complete Business Identity in one place, with free tools, resources, and built-in privacy from day one. Get more at northwestregisteredagent.com/SAASTRFREE
Sherif Mansour has spent 17 years at Atlassian and now runs AI across their entire product portfolio, including 20+ apps and Rovo, their AI platform used by over 5 million people every month. At SaaStr Annual, he shared the three biggest tensions his teams have wrestled with building AI into some of the most widely used SaaS products in the world, and why the answer to each one is almost never either/or. First: chat vs. dedicated UI. Atlassian built chat into every app not because they thought it was the end state, but because it became their best product discovery tool. Like the command line before Windows, chat is the universal interface. The patterns customers typed in chat told them exactly what features to build next. Second: reimagine vs. bolt on. The conventional wisdom says don't bolt AI onto existing workflows. Atlassian did exactly that, and it worked. They also landed on a principle that now applies across their entire portfolio: anything a human can do in your product, an agent should be able to do too. Third: the 10X AI builder myth. Atlassian ran experiments trying to make everyone an AI builder. What they found: everyone was rowing, no one was steering. They've since flipped their hiring pyramid, bringing in more juniors and seniors and fewer people in the middle, and their data shows juniors are nearly 2x more likely to experiment with AI tools than their senior counterparts. This one is packed with real examples, real data, and real lessons from shipping at scale. This episode is brought to you by Northwest Registered Agent. Starting a business can get expensive fast. Website here. Email somewhere else. Business phone with another provider... Northwest Registered Agent gives you a complete Business Identity in one place, with free tools, resources, and built-in privacy from day one. Get more at northwestregisteredagent.com/SAASTRFREE
We ran 22,000 paid SaaStr attendees from the last two years through enrichment and asked 10K to find the patterns. AI native job titles are up 4x in one year 63% of CMOs turned over in the last 16 months 75% of the AI native buyers coming to SaaStr are brand new to the community. The ICP is not who it was 18 months ago, and the data tells you who actually owns AI right now: CEOs, GTM leaders, and COOs. Then we get into agent pricing, which is chaotic. Salesforce, HubSpot, and Atlassian are all moving to charge extra for agent access. One of our vendors quoted us as much as $240K a year for what 10K currently does, about 40,000 API calls a day. So we watched what happens when an agent hits a wall: it immediately looks for workarounds, including mirroring data to Postgres and flagging those vendors as non-starters for anything new. The bigger question is whether we would have picked any of these tools at all if we knew this was coming. Meanwhile, Muse is free, already running competitor tracking, and one of our team members who has never touched 10K started spinning up his own agents in it this week. The bifurcation is real. We also cover collections, which nobody talks about but everyone suffers through. Since we made collections fully agentic, we went from 56% of invoices paid late to 8%. Average days overdue dropped from 17 to 6. Here's exactly what 10K does differently. --------------- This episode is brought to you by Northwest Registered Agent. Starting a business can get expensive fast. Website here. Email somewhere else. Business phone with another provider... Northwest Registered Agent gives you a complete Business Identity in one place, with free tools, resources, and built-in privacy from day one. Get more at northwestregisteredagent.com/SAASTRFREE
175% Growth, 120 Countries, Agents as Buyers: Stripe's CRO of AI on The New AI GTM Playbook The fastest-growing AI companies are rewriting every rule in the GTM playbook, and Stripe has a front-row seat to all of it. In this episode, Maia Josebachvili, Chief Revenue Officer of AI at Stripe, breaks down the four patterns she's seeing across the world's top AI companies, and what they mean for how you build, price, sell, and scale. The numbers alone are staggering: top AI companies grew 120% in 2025 and 175% in 2026. Lovable went from $100M to $400M in eight months. Cursor hit a $2B run rate in under two years. And 48% of revenue at top AI companies now comes from outside their home market. Maia covers: Why the fastest AI companies are in 42 countries on day one and 120 by year three How pricing is evolving from seats to usage to hybrid models, and why getting it wrong kills retention Why enterprise sales is now a year-one problem, not a year-five one What it actually means that agent traffic to Stripe's docs 10x'd in a single year How to build systems that move with your customer across PLG, enterprise, and agent-led motions If you're building an AI company or trying to compete with one, this is the data you need. This episode is brought to you by Northwest Registered Agent. Starting a business can get expensive fast. Website here. Email somewhere else. Business phone with another provider… Northwest Registered Agent gives you a complete Business Identity in one place — with free tools, resources, and built-in privacy from day one. Get more at northwestregisteredagent.com/SAASTRFREE