The landscape of sales and Go-To-Market (GTM) strategies is undergoing a profound transformation, driven by the rapid advancement of artificial intelligence. This shift was a central theme in a recent discussion between Jason Lemkin, founder of SaaStr and an early investor in the AI-powered unicorn Owner.com, and Kyle Norton, CRO at Owner.com. Owner.com, an AI solution, is revolutionizing how small restaurants manage their business, and Norton has been instrumental in propelling the company to nearly $100 million ARR in just a few years, with growth accelerating at scale.

Both Lemkin and Norton have extensively leveraged AI agents in their operations. Kyle Norton now oversees a 100-plus human sales team infused with AI, while Jason Lemkin and Amelia, SaaStr's Chief AI Officer, have deployed over 20 AI agents. Their experiences offer critical insights into the present and future of AI in GTM, highlighting a new era where AI agents are not just tools but integral team members.

Top 10 Takeaways on AI in Sales and GTM

  1. AI Agents Surpass Average Performers: AI agents are now more effective than mid-tier Account Executives (AEs) and Sales Development Representatives (SDRs). While they may not outperform the very best, their superior performance compared to average is enough to fundamentally reshape GTM team structures.
  2. Executive-Led Deployment is Crucial: CROs or CMOs who haven't personally trained and deployed at least one AI agent risk becoming obsolete. This isn't a task for agencies or consultants; it requires direct, hands-on involvement, typically 30 days of dedicated work.
  3. Focus on Few Tools, Go Deep: A common executive mistake is evaluating 8-10 vendors. It's impossible to effectively train 10 agents. Instead, choose two—one incumbent and one promising startup—and commit to deep deployment and training.
  4. Salesforce's Renewed Importance: Salesforce is making a comeback, not solely due to its "Agent Force" but because it serves as the essential hub for managing and resolving conflicts among 20 or more autonomously operating AI agents.
  5. The "Middle Ground" is Gone: The era of achieving lifestyle balance while exceeding quotas (common in 2021) is over. Professionals must now either work harder than ever for hyper-growth (10x, 5x, 5x, 5x rates) or join slower-growth companies (15-20%).
  6. Forward Deployed Engineers Over Features: Prioritize deployment support. Do not sign a contract until you've spoken directly with the engineer responsible for deploying your agent. The best vendor ensures successful production, not just a flashy demo.
  7. 30-Day Training is Non-Negotiable: Every AI agent requires a dedicated 30-day training period. This involves daily data uploads, output reviews, error correction, and iteration. Agents perceived as "not working" are typically those that were never properly trained.
  8. Address Core Pain Points First: Identify the most frustrating customer journey issues (e.g., website experience, question resolution) and deploy AI to fix those first.
  9. AI-Infused Teams are 3x More Productive: Kyle Norton's team at Owner.com achieves three times the revenue per AE compared to any previous team he's managed. This doesn't necessarily mean fewer reps, but rather higher quotas and continued hiring for growth.
  10. The Rise of the $250K SDR: Elite, genuinely 5-10x more productive sales professionals will command 2-3 times their previous salaries, but they will also be expected to deliver 10 times the output.

The Backstory: Why SaaStr Went All-In on Agents

Jason Lemkin's decision to go "all-in" on AI agents stemmed from frustration. He recounted an instance where two highly paid salespeople ghosted before SaaStr's biggest event. "I am done paying an SDR $150,000 a year or an AE $300,000 a year for basically inbound, spoonfed leads and renewals—and then having them quit on me," Lemkin stated. This experience, despite his loyalty and fair compensation practices, pushed SaaStr to seek an alternative.

Starting in May with one AI agent, SaaStr now operates over 20 agents in production, generating over $1 million in revenue. Lemkin's stark realization: "Our AI agents are better than a mid-pack AE or SDR. Not better than the best. But better than the 50th percentile person I’ve worked with over my career. And that changes everything."

The New Reality: Mid-Pack Sales Execs Are in Terminal Decline

Lemkin is blunt about the future for average GTM professionals: "If you’re a mid-pack GTM professional who doesn’t want to work harder and smarter than a year ago, these jobs are in terminal decline." He cited SaaStr's experience sending 70,000 hyper-personalized emails for SaaStr London using AI agents, which outperformed the 7,000 emails sent by humans in terms of volume and slightly better quality. In contrast, a highly paid SDR's reluctance to follow up on a lead highlighted the efficiency gap: "The agent? The agent doesn’t argue. The agent just follows up."

But We're Still in Inning One

Despite current advancements, Lemkin emphasizes that AI in GTM is just beginning. Today's "hyper-personalization" often means only a few dynamic fields in an email. The true potential lies in AI leveraging comprehensive data—every competitor, website visit, brand interaction, and adjacent tool in a prospect's stack—to craft emails as compelling as those meticulously created by top human minds. "When it is? Buy that product immediately," he advises.

The Real Reason Agent Deployments Fail

Early AI SDR failures in 2024 were largely due to immature LLM technology. However, with advancements like Claude 4, the underlying technology is now robust. Today, agents fail because "people don’t roll up their sleeves and train them." Lemkin shared an anecdote where a RevOps tool was deemed ineffective because an AE hadn't bothered to link their account, revealing a lack of engagement rather than a tool malfunction. The lesson: without dedicated training, even the best tools will appear to fail.

The 30-Day Rule for AI Agent Training

Effective AI agent deployment requires weeks of rigorous training. This process breaks down into three phases:

  • Day 1-7: Ingestion
    • Upload prospectuses, documentation, and connect to databases or websites.
    • The agent generates a list of sample outputs (10-30 questions).
  • Day 8-21: Iteration
    • Review every single output daily.
    • Correct inaccuracies (e.g., "Owner is great for 100-chain high-end restaurants" becomes "Owner scales much more now, but our core audience is single-location restaurants with significant to-go business").
    • The agent learns and improves daily.
  • Day 22-30: Production
    • Hallucinations become a minor issue.
    • The agent is ready to scale.

Neglecting this process will lead to the false conclusion that the agent doesn't work.

Do It Yourself, Dude: The Executive Imperative

Lemkin's strongest advice for CROs and CMOs is unequivocal: "If you don’t roll up your sleeves in the age of AI and AI GTM, you will become obsolete." He stresses that AI GTM is not yet an agency game; executives must be the "agency" for now, at least for the first deployment. "If you haven’t trained an agent yourself, you have no idea what you’re talking about. Literally. You will be utterly ignorant in the age of AI." Lemkin himself trained SaaStr's first agent daily for a month to understand the process before delegating further.

How to Pick Your First AI Agent

Selecting and deploying your first AI agent involves a strategic approach:

  1. Identify a Medium-to-High Ranking Problem: Choose a tool that addresses a significant pain point, whether it's AI SDR, RevOps, or support. Pick something you're passionate about or that currently frustrates your customer journey.
  2. Find the Right Vendor Partner: Prioritize deployment support. "Don’t sign the contract until you talk to your Forward Deployed Engineer (FDE)." Lemkin advises asking directly to speak with the person who will deploy the agent, even before a demo. A vendor unwilling to connect you with their deployment team is a red flag.
  3. Budget for $50-100K, Not Headcount: The initial investment is for the tool, not a reduction in headcount. Secure a budget of $50K-$100K. Deploy it yourself, prove the ROI with data (e.g., "We just sent 70,000 automated emails... It generated 15% of the revenue for SaaStr London"), and then approach your CFO for more budget.

The Two-Vendor Bakeoff

The old practice of conducting 8-10 vendor bakeoffs is no longer feasible. With each agent requiring 30 days of training, evaluating so many is impossible. "Do two," Lemkin advises. Select one incumbent option (e.g., Intercom's Finn, Zendesk's AI, Salesforce's Agent Force) and one promising startup. Seek references via email, as leading figures like Kyle Norton receive numerous requests and can provide honest, concise feedback. The key is successful production, as leading agents are rapidly converging in features.

The Lowest Hanging Fruit: Fix Your Inbound

The easiest immediate win for AI deployment is enhancing inbound processes. Prospects should receive instant answers to questions, and there's no longer a need for a 21-year-old SDR to qualify leads before connecting with an AE. SaaStr's digital assistant, Amelia, provides instant qualification and answers, eliminating waiting times and scheduling hurdles. High-friction sales processes are obsolete; AI can score leads more effectively than humans today.

Why Salesforce is Back

Both Kyle Norton and Jason Lemkin are leaning into Salesforce more than ever. With multiple autonomous AI agents working across different segments, a central hub is essential for data consolidation and conflict resolution. Salesforce serves this purpose. While agents may extract the majority of the value, Salesforce's native integration proves beneficial, making Agent Force a critical investment for the company, despite requiring more setup work than competitors.

The 3x Productivity Question

Kyle Norton's AI-infused team at Owner.com demonstrates 3x higher productivity per AE than any team he's previously managed. This doesn't necessarily mean fewer hires, but rather higher quotas and continued recruitment for growth. AI introduces leverage into sales, a historically leverage-averse field. This means genuinely elite, 5-10x more productive professionals can expect 2-3 times their previous compensation, provided they deliver 10x the output.

Triple Triple Double Double Isn't Enough Anymore. At Least Not for VCs.

While "triple triple double double" growth (e.g., $2M to $6M to $18M ARR) can still build a multi-billion dollar company, it's no longer enough to attract most venture capital. What once appealed to 50% of VCs now only interests about 10%. With venture capital flowing to fewer companies, startups must make their existing capital last, recognize that top talent gravitates towards the fastest-growing companies, and be brutally honest about their fundability. Tools like saastr.ai/aivc can provide data-driven fundability analysis.

Pick Your Path: The Bifurcation of Professional Life

The professional landscape is bifurcating into two distinct paths:

  1. Work Harder Than Ever: Achieving venture-scale outcomes (10x, 5x, 5x, 5x growth) demands unprecedented effort. Both Lemkin and Norton attest to working harder than ever, even with tailwinds and strong teams.
  2. Join Something Slow-Growth: Companies growing 10-20% annually still offer good compensation and manageable pressure. This path allows for a more traditional work-life balance, logging off earlier without constant work-related thoughts.

The "magical middle" of 2020-2023—where lifestyle, remote work, and exceeding quotas with 20-30 hours a week were possible—no longer exists. Even the fastest-growing startups are now lean, office-centric, or high-intensity. Professionals must be honest about their desires and choose their path accordingly.

Top 5 Mistakes Execs Make with AI GTM Agents

  1. Running 8-10 Vendor Bakeoffs: It's impossible to train 10 agents effectively. Executives should pick an incumbent option and one hot startup, then conduct real deployments of both.
  2. Outsourcing Deployment to Agencies or Consultants: Current AI GTM agencies lack the deep business understanding needed to train agents effectively. For now, executives must be hands-on.
  3. Buying a Tool and Shelving It Without 30 Days of Training: Agents perceived as "not working" are simply untrained. Daily data uploads, output reviews, and error correction are essential until reliable operation is achieved.
  4. Not Talking to the Forward Deployed Engineer Before Signing: The best vendor is defined by its ability to get you into production, not just its features. Direct communication with the deployment team is critical before committing.
  5. Giving AI Tools to Individual Reps to "Figure Out on Their Own": The old paradigm of letting reps manage their own cadences doesn't work with AI agents. A centralized, "nerdy GTM person" at the top of the stack is needed to manage the entire system.

Quotable Moments

Jason Lemkin: "I am done paying an SDR $150,000 a year for basically reaching out with mediocre emails to leads that are already high qualify—and then having them quit on me."

Jason Lemkin: "Our AI agents are better than a mid-pack SDR, and in part, AE. Better than the 50th percentile people I’ve worked with over my career. And so we just don’t need them."

Jason Lemkin: "If you don’t roll up your sleeves in the age of AI and AI GTM, you will become obsolete."

Kyle Norton, CRO Owner: "We’ve got nine high-impact production use cases of AI. Our booked revenue per dollar spent, on a per-AE basis, is 3x any team I’ve ever managed before."

Kyle Norton, CRO Owner: "We built our cold outbound email program infrastructure in three weeks. Everybody else told us that was a multi-month project."

Kyle Norton, CRO Owner: "Even at Owner, even with all the tailwinds, even with an awesome team—I’m still working the hardest I’ve ever worked."