Jason Lemkin shares how AI agents replace 10 SDRs with 20 agents, revolutionizing GTM. Learn what this means for your startup's sales strategy and team struc...
How AI Agents Are Transforming Go-to-Market Strategy in 2025
Key Insights
- AI is eliminating traditional SDR roles – Cadence-based email SDRs will be 90% replaced within 12 months
- SaaStr replaced 10 GTM employees with 1.2 humans + 20 AI agents – Achieving similar revenue with exponentially fewer staff
- Best sales practices now require agent training and orchestration – Not just buying software, but building expertise in deployment
- The GTM market is polarizing dramatically – Fastest-growing companies win because they're overwhelmed with demand; slowest-growing need ruthless efficiency
- Customer success and support are already transformed – 50-80% of support tickets are now handled by AI across most vendors
- High performers will thrive with AI as a superpower – Average performers will simply become more average
The Future of Sales in the AI Era
What's Changing: The End of Traditional Sales Roles
The most profound shift happening in go-to-market strategy isn't gradual—it's accelerating rapidly. At SaaStr, we made a bold decision after our top two Account Executives quit on the same day during our annual conference. Rather than hiring replacements, we decided to experiment: what if we could scale our sales function using AI agents instead?
The results have been striking. Our previous team had 10 dedicated GTM professionals handling sponsorships, ticket sales, and customer outreach. Today, just 1.2 humans (with 0.2 of that being our Chief AI Officer, Amelia, who dedicates 20% of her time to agent management) manages 20 AI agents. The business is producing similar results to when we had a full team—but these agents work 24/7, including nights, weekends, and holidays.
Walk into our SaaStr office today, and you'll see something unusual: ten desks that once belonged to our GTM team now have nameplates for our AI agents. We have Replit's Repli, Qualified's Quali, Artisan's Arti, and others. These aren't experiments anymore—they're core to how we operate.
The specific roles being eliminated aren't controversial in hindsight. Email-based SDRs who send cadences through tools like Outreach or Salesloft? That role will be 90% gone within 12 months. Business Development Representatives who validate inbound leads and manage "contact me" requests? Also largely obsolete. There's no reason a customer should wait days for a 21-year-old SDR to respond when an AI agent can instantly qualify them and schedule a meeting—often without the customer even realizing they've been qualified by an agent.
However, the transformation goes deeper than simple job displacement. What's actually happening is a restructuring of who creates value in go-to-market. The 20% of sales professionals who will genuinely thrive are those who evolve into agent orchestrators—people who understand how to train, deploy, and continuously improve AI agents to handle complex sales workflows.
The New Sales Hierarchy: What Actually Creates Value
Here's the uncomfortable truth that most GTM leaders aren't discussing openly: if your primary differentiator is that you're "good with people," you're operating with an obsolete skillset. That might sound harsh, but consider what we've discovered after running hundreds of thousands of AI-generated emails and conversations.
When we looked at high-performing SDRs versus AI agents we'd trained on their best practices, the gap narrowed significantly once we invested in proper training. But more surprisingly, we found that even mediocre human salespeople were sending low-quality emails. When we fed our best templates and closing strategies into our AI agents, they weren't just competitive—they were often superior to average human performers.
The data tells a clear story: AI will enhance top performers into superhumans. These individuals will manage 10+ agents instead of managing direct reports. A premium SDR salary of $250,000 becomes someone who orchestrates a whole team of agents and earns that compensation for their mastery. But the middle 70-80%? They'll find themselves competing against agents trained on best practices, and they'll likely lose that competition.
What's required to survive and thrive in this new world?
For individual contributors: Master the specific tools your organization deploys. If your company chooses Qualified, become the expert on Qualified. Learn how to train the agent, understand its capabilities and limitations, and position yourself as someone who can extract maximum value. You'll become exponentially more productive—and more valuable.
For managers: You must directly purchase, deploy, and train agents yourself. You cannot delegate this to agencies or consultants. Why? Because these AI systems require intimate knowledge of your business, your sales processes, and your best practices to function effectively. The Forward Deployed Engineer (FDE) or Solutions Engineer who helps you is a partner, not a replacement for your own deep engagement. The managers who do this work will become Chief Orchestration Officers, managing the people-to-agent ratio and driving 10x productivity gains.
For leaders: Be brutally honest about what's changing. We made the decision after those two SRs quit: we would no longer hire junior GTM staff. Instead, we'd invest in building our AI infrastructure while retaining only senior people who could orchestrate agents and own customer relationships. This isn't cruel; it's honest about where the market is heading.
The Training Imperative: Why Most AI Implementations Fail
One of the most critical insights we've learned is also the most commonly ignored: these agents don't work without serious training and orchestration.
When we spoke with the Chief AI Officer of a publicly traded B2B company worth over $10 billion, they'd purchased an AI SDR tool and expected their junior team to figure it out independently. When we asked if anyone on their team of 20 had actually trained an agent themselves, the answer was a resounding no. They expected magic—just activate the tool and watch sales happen.
That's not how this works.
Training an AI agent requires understanding what we call the "intake-orchestration-training" framework. While these terms sound technical, the process is actually familiar to anyone who's done B2B marketing or sales at scale:
Intake: Upload your training materials—website content, product documentation, sales scripts, training documents, case studies, etc. The system processes and structures this data.
Orchestration: Set up workflows that determine what the agent does, when, and to whom. This might involve email sequences, web chat interactions, calendar integrations, or CRM connections.
Training: This is the ongoing work. You'll start with a few hours of initial training, then commit to daily refinement. When an agent hallucinates (provides incorrect information), you correct it. When an email template performs better than another, you reinforce it. After 30 days of this daily investment, your agent moves from "fundamentally broken" to "pretty good."
We ran this process across multiple vendors. When we trained agents on our sponsorship process with Qualified, we saw 70% response rates from prospects who expressed interest but hadn't moved forward. When we used Artisan for outbound campaigns to previous attendees, we sent 60,000 emails and achieved strong conversion metrics. When we deployed Salesforce's Agentforce, we trained three separate agents over months, then realized we could compress all that learning into a single prompt that got us 80% of the way there in one day.
The common thread: every successful deployment required direct, hands-on involvement from someone who understood both the business AND the tool. This person becomes your competitive advantage.
Who Actually Needs to Make This Investment
Here's the reality check: these agents cost money. The entry price for serious AI GTM orchestration runs $50,000-$70,000+ annually, often with additional services costs. Individual contributor SDRs won't have access to this budget. That's a VP, director, or founder decision.
But if you make that investment, if you personally train your agents, if you understand exactly how they work and continue to optimize them, you become phenomenally hireable. Organizations are desperate for people who can actually deploy this technology. Most are stuck in "panic mode." You'll have negotiating power unlike anything in traditional GTM roles.
The Paradox of Demand: Why Growth is Fracturing
Here's something most people aren't discussing: the go-to-market landscape is becoming radically polarized.
At the top, AI-native companies like Vercel, Replit, Eleven Labs, and emerging competitors have an embarrassment of riches. They're overwhelmed with inbound demand. One competitor even lost a seven-figure deal because their sales team simply never called—they were drowning in opportunities. These companies are hiring more people (to manage explosive growth) despite becoming more efficient.
At the bottom, traditional SaaS companies are under margin pressure. They need ruthless efficiency. They're investing heavily in AI to do more with less.
But in the middle—the comfortable zone most companies occupy—there's a dangerous squeeze happening. Companies aren't experiencing the viral growth that makes AI-native efficiency optional, and they're not under enough margin pressure to force the optimization conversation. So they're drifting.
Internally at SaaStr, we're already at capacity. We can manage 20 agents, but a 21st agent might tip Amelia's workload into unsustainability. We've hit the orchestration ceiling with current tools and current human capacity. And we're not unique in this observation. The vendors building these agents are struggling with customer success because deployment is hard. Salesforce has invested massive resources into training their own professional services team for exactly this reason.
What Remains Constant: The Fundamentals
Here's what won't change: the businesses that win will still be built on genuine customer value, deep product understanding, and authentic relationship building at the decision-maker level.
The difference is in scale and efficiency. A talented human salesperson who deeply understands their product and the customer's business will always close bigger deals than an AI agent. But that person will also close 3-4x more deals because the agents handle everything else. They focus their human judgment and expertise on moments that matter.
Cold calling, for instance, remains a specialized skill that very few people execute well. If you're exceptional at cold calling, keep doing it—AI isn't there yet. But most SaaS companies aren't actually winning deals through cold calling. They're winning through combination approaches: inbound demand, product-led growth, community engagement, and targeted outbound. The AI revolution is most disruptive in the high-volume, high-touch workflows that exist between initial contact and qualified opportunity.
Practical Steps for Different Roles
If you're an individual contributor (SDR, BDR, AE):
Embrace the tools your organization deploys. Yes, it might feel threatening. But if you become the person who gets the most out of these agents, you're not at risk—you're differentiated. Someone at your company will be running 10 agents. Make sure it's you. If it's not you, you're vulnerable.
You're also seeing something important: the bar for "good enough" in sales is rising. Mediocrity gets eliminated. Excellence gets amplified. This is actually better for talented people. Your effort compounds more dramatically.
If you're a manager (VP Sales, Chief Revenue Officer):
You must become an AI operator yourself. Purchase an agent. Train it. Deploy it. Live with it for 30 days. Make it your daily focus. This is non-negotiable. You cannot outsource your understanding of how these tools work. Your team is watching to see if you're serious about the transformation.
When hiring for GTM orchestration roles, look for people who've actually done this work. Not consultants. Not people with AI certification. People who've shipped agents into production and optimized them weekly. They're rare, and they're getting multiple job offers. Honestly, if you're strong at this, you can name your salary in 2025.
If you're a founder:
You have both an obligation and an opportunity. The obligation: be honest with your team about what's changing. Don't hide behind "AI enhances human potential" platitudes. Say "Some roles we used to hire for no longer make sense. Here's how we're evolving, and here's where we see opportunity for people who are willing to learn."
The opportunity: You're in the position to make bets on where AI can create competitive advantage. Try the secret-mode customer journey test we mentioned. Go through your entire customer experience as if you were a stranger. Note everything that's broken, slow, or frustrating. Pick the single most painful thing. Find an AI agent tool that might solve it. Train that agent. Deploy it. Watch it work 24/7 on something that humans actively avoid because it's low-value work.
You'll get excited. You'll see the potential. And you'll start building the future of your GTM function instead of defending the past of it.
The Larger Shift: What's Actually Happening
What we're witnessing isn't just a change in tools or tactics. It's a fundamental restructuring of go-to-market economics.
When you can have 1.2 humans manage 20 agents that operate 24/7 and produce results equivalent to 10 humans working traditional hours, you've changed the economics of scale. You've made it possible for companies with small teams to operate like companies with much larger organizations. You've made geographic location irrelevant. You've made hiring for "cultural fit" versus "core competency" a much lower priority.
This accelerates winner-take-most dynamics. Efficient companies get faster. Inefficient companies get slower. The gap widens. Over 5-10 years, this compounds into radically different outcomes.
But here's the nuance: this doesn't mean GTM is shrinking overall. The companies that win in an AI-driven world still need humans—just different humans. They need orchestrators, not cadence managers. They need people who understand AI's capabilities and limitations. They need humans who can recognize when a customer relationship requires authentic human judgment (which AI agents generally still lack) versus situations where AI can execute faster and more reliably.
Preparing for 2025: Your Action Plan
If you're in go-to-market—whether as an individual contributor, manager, or leader—here are the concrete steps to take over the next 90 days:
Choose a tool and commit to it. Don't spend three months comparing six vendors. Pick one—Qualified, Artisan, Salesforce Agentforce, or another—based primarily on which vendor will help you the most. Work with their Forward Deployed Engineer. Invest the relationship.
Solve one painful problem. Customer support? Sales cadence? Inbound qualification? Pick the one that causes your team the most frustration. Build an agent to handle it.
Train deeply. Spend 1-2 hours daily for 30 days. Upload materials. Test queries. Correct hallucinations. Document what works and what doesn't. This is not optional busywork—it's how you develop competitive advantage.
Measure obsessively. Volume, quality, conversion, customer satisfaction, internal team feedback. What's working? What's not? Where are your agents failing?
Iterate relentlessly. After 30 days, you'll have enough data to improve significantly. After 90 days, your agent should be genuinely useful. After 6 months, it might be better than your average human at that task.
Expand thoughtfully. Once you've successfully deployed agent #1, start agent #2. But don't overextend. Orchestration is the bottleneck, not the agents themselves.
The companies and individuals who do this work in January 2025 will be dramatically ahead by June 2025. The landscape is moving at incredible velocity. Your competitive window is open, but it closes relatively quickly.
Conclusion
The go-to-market revolution driven by AI isn't a future event—it's happening now. Companies are making bet-the-business decisions about AI agent deployment. Individuals are repositioning themselves as orchestrators rather than executors. The market is sorting rapidly into companies that are forward-looking and companies that are defending yesterday's playbook.
The choice before you is binary: Move forward aggressively and become the expert in your organization who can harness AI agents to drive growth, or watch as your expertise becomes incrementally less valuable as average tools get better and better.
This is likely your best startup yet—if you're willing to transform the one you're already running. Don't jump to chase the next shiny opportunity. Instead, bring AI native innovation to your current business. Train an agent. See what becomes possible. Then build from there.
The magic moment is now. The frontier is open. Your move is next.
원문출처: https://www.youtube.com/watch?v=I-R1bc1rlFs
powered by osmu.app