The sales and GTM sessions at SaaStr AI 2026 in San Mateo was all about where to take agents next.
No one was still debating whether to put agents in the revenue org. They had already done it, and they brought the war stories. What broke, and what they would do differently next time.
The lineup behind these Top 12 lessons:
- Eleanor Dorfman (Head of Industries, Anthropic) on rebuilding the sales org so 54% of new enterprise logos close self-serve
- Grant Lee (Co-Founder and CEO, Gamma) on hitting $100M ARR with almost no sales team
- Kyle Norton (CRO, Owner.com) on AI-native GTM and the five calls every CRO has to get right
- Jeanne DeWitt Grosser (COO, Vercel) on the lead agent that took a 10-person function down to 1
- Kody (Sales, Replit) on the data showing rep-level AI usage predicts quota attainment
- Adam Alfano (President, Salesforce) and Eitan Saban (Head of North America Mid Market Sales, PayPal) on selling SMB with agents
- Sam Blond (Co-Founder and CEO, Monaco) on comp, headcount, and margin math when agents deliver the outcome
- Maia Josebachvili (GM of Enterprise Product, Stripe) on the four patterns behind the fastest-growing AI companies
Pretty good line-up.
1. When demand surges, open a self-serve path instead of just hiring reps
Eleanor Dorfman, Head of Industries at Anthropic, drew one of the biggest crowds of the event, and the content earned the room. She walked through what her team did when a new Claude release sent enterprise demand vertical. The obvious move was to hire reps three to five times faster. She argued you cannot do that without wrecking the buying experience, so Anthropic rebuilt the enterprise motion around AI instead.
Four months after the rebuild, 54% of new enterprise logos were closing through self-serve. Not trials, and not small accounts. Real enterprise logos, real ACV, real contract terms, real invoicing, with no rep gating the front door. The reps who used to run those deals got pointed at the accounts where a human actually changes the outcome.
This sits in direct tension with the next lesson, and both are right. Gamma says most companies add sales too late. Anthropic says that when demand spikes, the instinct to throw reps at it is often the wrong first move. The way to hold both is to treat human selling as expensive and scarce, spend it only where it moves the deal, and let everyone else buy without waiting on a calendar.
→ Top Takeaway: When demand spikes, do not just hire reps. Build a real self-serve enterprise path so buyers who do not need a human never wait for one, and reserve your reps for the deals that actually need them.
2. Add sales earlier than feels necessary, even when inbound is carrying you
Gamma is the single best argument in B2B for skipping a sales team, and Grant Lee stood on stage and told founders not to skip it.
Gamma hit $100M ARR with roughly 50 people. Profitably. 50 million users, 600,000 paying subscribers, and almost all of it driven by word of mouth rather than a sales org. If any company earned the right to say “we don’t need sales,” it was this one.
His biggest regret was waiting too long to add it.
Inbound clearly works. Gamma proves it works better than almost anyone thought possible. Even so, a world-class inbound motion leaves enterprise deals, expansion revenue, and larger accounts sitting on the table, and you only find out how much after you finally hire the people who go get them. If the company with the strongest excuse to wait wishes it had moved sooner, most founders riding inbound are later than they think.
→ Top Takeaway: If a $100M inbound machine regrets waiting, you are later on sales than you think. Hire ahead of the pain, not after it shows up in the forecast.
3. The new bar for a rep is 20x their OTE
Kyle Norton, CRO at Owner.com, put up the most concrete rep economics of the event. Owner is approaching $100M ARR selling roughly $10K ACV software to independent restaurants, the kind of small businesses that supposedly don’t buy software.
The numbers on his AI-native team:
- $2M+ in ARR per rep per year, as the average, not the top performer
- 20x close-won to OTE, meaning a $150K rep is bringing in multiples of their comp
- $100K+ in closed-won per outbound BDR per month, closed revenue, not pipeline
- 4x the ARR per rep of their direct SMB competitors
The takeaway for every revenue leader is a reset on what “good” looks like. If your reps are running at 3x or 4x their comp and you think that’s healthy, the ceiling just moved. AI-augmented reps at a well-run org are producing at a level that makes the old benchmarks look outdated.
→ Top Takeaway: Reset your rep benchmarks. In an AI-native org, 3x comp is no longer the ceiling, it is the floor.
4. Point agents at the leads no human was ever going to call
The PayPal and Salesforce session, with Adam Alfano from Salesforce and Eitan Saban from PayPal, delivered the cleanest ROI story of the week.
PayPal put Agentforce on roughly 8,000 leads a month that no human was going to touch. These were not the good leads. These were the accounts sitting at the bottom of the pile, the ones the team had already triaged out. Conversions on that pool jumped 50%, and they saw it inside the first few months.
That is the fastest payback available in AI sales right now, and most companies are ignoring it. Everyone wants to point agents at the top of the funnel where reps already live. The bigger win is the pipeline you already gave up on. You are not taking work away from a human. You are recovering revenue that was going to zero.
→ Top Takeaway: The fastest AI ROI is not better leads, it is the pipeline you already gave up on. Point an agent at your dead leads this quarter.
5. Rebuild comp before agents break it
Sam Blond, co-founder and CEO of Monaco and former CRO at Brex, ran the first session at SaaStr to seriously work through the comp math of AI-delivered outcomes.
The question he forced into the open: if an agent books the meeting, qualifies the lead, and writes the follow-up, who gets paid? When AI qualifies the leads that humans close, and humans start deals that AI finishes, individual attribution stops meaning anything. The whole apparatus of sourced-versus-closed, SDR-to-AE handoff credit, and per-rep quota was built for a world where a person did every step.
Most orgs are going to hit this the hard way, mid-year, when a comp plan designed in 2025 collides with a team where agents do half the pipeline work. Get ahead of it. The right question is no longer “how do I pay this rep for this deal.” It is “am I optimizing for individual rep performance or total enterprise value,” and those two answers are diverging fast.
→ Top Takeaway: Rebuild the comp plan before agents are doing half the work under a plan that assumes humans do all of it.
6. Monetize earlier and go global by default
Maia Josebachvili, GM of Enterprise Product at Stripe, works with the fastest-growing AI companies in the world, which gives her a clear read on the patterns that separate them.
Two stood out for sales leaders. The fastest movers monetize far earlier than the previous generation did, and they are global from day one rather than treating international as a phase-two project. The old sequence was build, scale users, then monetize, then expand geographically. The companies pulling away are compressing all of that.
She framed it against her own history. At her first company, Urban Escapes, the original checkout flow told customers to mail a check to her apartment, and people did it. She wasn’t being nostalgic. The tolerance for a clunky, delayed, US-only monetization motion is gone. If you are waiting to charge and waiting to go global, someone built to do both on day one is already in your market.
→ Top Takeaway: Charge sooner and sell globally from day one. The grace period for waiting on either is gone.
7. Agents are becoming buyers, not just tools
Maia Josebachvili’s other pattern from the Stripe session is the one most sales teams have not priced in yet. Stripe sees the actual transactions across the fastest-growing AI companies, and Josebachvili is watching agents start to initiate and complete purchases inside real commerce flows. Not a human clicking buy after an agent does the research. The agent itself transacting.
That means your checkout, your pricing page, and your buying process now have a non-human user. An agent evaluating you does not read your landing page copy or respond to a well-timed follow-up. It checks whether your product is callable, whether your pricing is legible to software, and whether it can complete a purchase without a human step in the middle.
For sales, this is a new surface to design for. The teams that make it easy for an agent to evaluate, quote, and buy will capture demand that never shows up as a lead in the CRM. The teams that assume every buyer is a person will quietly lose deals they never knew were in play.
→ Top Takeaway: Treat agents as a real buyer segment. Make your product callable, your pricing machine-legible, and your purchase path completable without a human in the loop, because some of your next buyers will be software.
8. Centralize your AI or the gains never compound
Back to Kyle Norton, because his second big idea is one most companies are getting wrong right now. He laid out an AI sophistication ladder, and the trap is sitting on the wrong rung and thinking you have arrived.
Level 1 is individual reps and RevOps building their own custom GPTs and prompts. It feels like progress. It is where most B2B companies are stuck. The problem is that every experiment stays local, the great ideas never scale past the rep who built them, and you get a thousand small wins that never add up.
The compounding leverage lives at Level 3: centralized infrastructure, shared skills, and a common context library that every rep draws from. His hard advice was to stop letting reps run their own agents in isolation and stand up a GTM engineering function that builds for the whole team. The gap between the companies doing this and everyone else is widening exponentially, not linearly.
→ Top Takeaway: Stop letting reps run private agents in isolation. Centralize the infrastructure and shared context, or the gains never compound.
9. Buy AI sales tools like a VC who has seen 200 pitches
Before founding Monaco, Sam Blond evaluated hundreds of AI sales startups from inside Founders Fund. His read: most of them are dead on arrival, and the pattern separating the real ones from the GPT wrappers is clear once you have seen enough.
That is a buying lesson as much as a founding one. Every revenue leader is now getting pitched a dozen AI sales tools a quarter, and the demos all look the same. The discipline is knowing which questions actually predict whether a tool will work in your org versus which ones just produce a polished demo. Ask what happens on the deals that do not fit the happy path. Ask what it writes to your system of record and how it fails. The vendors who cannot answer are the ones whose product does not really work, and they will happily take twelve months of your team’s attention before you find out.
→ Top Takeaway: Pressure-test every AI sales vendor on the unhappy path. If they cannot explain how it fails and what it writes to your CRM, it does not really work.
10. One good agent can replace your entire SDR layer
Jeanne DeWitt Grosser, COO of Vercel, put up the most concrete headcount number of the event. Vercel built a lead agent that took a function which used to need a team of 10 down to a single person, and it returned 32x on what they put into it.
This is the part of the org that compresses first. The volume layer, the SDRs and researchers doing repeatable qualification and outreach, is exactly the work an agent does well. Grosser’s team did not sprinkle AI on top of the existing motion. They rebuilt the workflow around the agent and kept one human to steer it. The closers stayed. The layer feeding them got automated.
The warning sits inside the win. A 32x number only shows up when you redesign the process, not when you bolt an agent onto the old one. The companies attaching an agent to a 10-person SDR team get a small lift. The ones rebuilding the function around the agent get Vercel’s number.
→ Top Takeaway: The volume layer of your sales org compresses first. Rebuild the SDR function around one good agent instead of hiring the next five reps.
11. Your reps’ own AI usage predicts whether they hit their number
Replit sent its whole senior team to the event, CEO, CRO, and President, which tells you how seriously they take this. The most useful thing they shared was a dataset. Kody, who runs sales at Replit, correlated how much each rep used Replit internally against that rep’s quota attainment. The pattern was clear: the reps leaning hardest on the AI were the ones hitting their numbers.
Replit is approaching $1B in revenue in one of the most competitive categories in software, and their read is that the only way to survive there is to amplify each person’s output by 10x or more with agents. Rep-level AI usage is not a soft skill you hope for. It is a leading indicator of performance you can actually measure.
For a sales leader, that changes what you track. You already watch activity, pipeline, and attainment. Add AI usage to the board, because it shows up in the number before the number does. It changes hiring too. A rep who is genuinely AI-fluent is not just faster to onboard, they are measurably more likely to carry quota.
→ Top Takeaway: Track each rep’s AI usage and treat it as a leading indicator of attainment. Hire for AI fluency, and coach the reps who are not in the tools before it shows up in their number.
12. The mediocre-rep tier is gone, and the top of the team is worth more
The thread that ran through nearly every revenue session was blunt, and Kyle Norton said it most directly. AI agents are already better than mid-pack AEs and SDRs. Not better than your best people. Better than average. And that is enough to change how you build the team.
The middle of the sales org, the reps who were fine but never great, is the part under real pressure. At the same time, the best AI-native reps become worth more, not less, because they are now running the output of what used to take three people. The prediction on stage was that the strongest AI-powered SDRs command well above today’s comp for far more output.
The hiring implication is immediate. Anyone coming into a revenue leadership seat has to demonstrate real, hands-on AI usage, not enthusiasm about it. If a candidate is not already in the tools, they cannot lead a team whose entire advantage now depends on them.
→ Top Takeaway: The middle of the team is the risk, not the top. Hire AI-native, invest in your best reps, and require hands-on AI use for every leadership seat.
More worth stealing
A handful of one-line takeaways that didn’t make the main 10 but are worth writing down.
- The AI-augmented rep day is 4 demos and 2 strategic follow-ups, with zero time on CRM hygiene or note-taking. Kyle Norton (Owner) is pushing his team to 70-80% revenue-generating time, against the 20-30% that’s normal at a traditional org.
- The $250K SDR is coming. Kyle Norton (Owner) expects the best AI-native SDRs to earn two to three times today’s comp for roughly ten times the output. The role gets more valuable, not less.
- Nobody has hit the top of the ladder yet. Kyle Norton (Owner) defines Level 4 as a recursively self-improving GTM system, and says he has not found a single B2B company actually operating there. The race is still open.
- The 50% conversion lift landed in about 14 weeks. Eitan Saban (PayPal) shipped the Agentforce results in a single quarter, not a multi-year transformation. Speed to proof is now measured in weeks.
- Manufacture one brand moment every 30 days. Sam Blond (Monaco) runs demand creation as a repeatable engine, the plane over San Francisco, the six-figure invitational, the busiest booth at the show. A plane is cheaper than a freeway billboard.
- The first 30 seconds have to feel like magic or inbound never starts. Grant Lee (Gamma) treats the opening moment of the product as the actual top of the funnel. No wow, no word of mouth, no pipeline.
- Founders have to become creators. Grant Lee (Gamma) puts distribution on the founder building in public, not on a marketing team waiting downstream. The audience is the first channel.
What this means for how you build a revenue team in the Age of AI
he sales org that wins from here is smaller, more senior, and AI-native. It points at pipeline humans used to write off, lets buyers who do not need a rep close on their own, and is ready for the buyers that turn out to be software rather than people. Comp and hiring have to be rebuilt to match, and the companies that rebuild first are the ones putting up the numbers Owner and Gamma showed on stage.
