We’ve been running AI GTM agents for almost a year now. We use Artisan for outbound, Qualified (now Salesforce) for inbound, Agentforce for Salesforce-native outreach, Delphi for Digital Jason. We’ve sent 20,000+ AI messages and generated over $2M in closed revenue from our AI SDR on its own (plus much more where AI SDR set up meetings for our humans).
And these tools are genuinely impressive. They’re saving real time. They’re generating real pipeline. They’re good.
But here’s what I keep coming back to:
They’re not even close to what Cursor and Replit have achieved in engineering.
And that gap? That gap is everything.
The Numbers Tell the Story
Let me put this in perspective.
Cursor went from $1M to $500M ARR in under two years. It hit $1B+ in annualized revenue by November 2025. It’s now valued at $29.3 billion. Half the Fortune 500 uses it. Developers report 20-55% faster delivery times. 72% of professional developers now use or plan to use AI coding assistants daily.
Replit crossed 22.5 million users and 500,000 businesses. Users created 2 million apps with AI in just six months. It went from $10M to $250M ARR in twelve months on AI-driven app building alone.
The AI SDR market? It’s growing rapidly, but it’s not that. But it could be.
What We’re Actually Running (And What It Actually Does)
Let me be real about what our AI GTM stack actually does:
Artisan (Outbound AI SDR):
- 19,326 messages sent over 6 months
- 3,221 emails/month from a single platform (vs. 75-285/month from a human SDR — that’s 11-43x volume)
- 6.67% overall response rate (vs. 2-4% industry average)
- Warm campaigns hit 12.13% positive response rates
- 10% of SaaStr AI London ticket revenue sold directly via AI SDR
- Booked a six-figure sponsor meeting on a Saturday at 6:02 PM
Qualified (Inbound AI SDR, now joining Salesforce):
- 668,591 total sessions since August
- 1,025 meaningful conversations
- 91 meetings booked
- $1,010,000 closed-won revenue in 90 days
- $2,500,000 currently in pipeline
- 71% of October’s closed-won deals came from AI-qualified inbound (historic average: 29-34%)
Agentforce (Salesforce-native AI SDR):
- Deployed just over a month ago at Dreamforce
- 72% open rate (vs. 0% when humans never sent the email)
- Highest response rate of all our AI platforms
- ~3,000 emails sent to previously ghosted leads
- Already closing deals from contacts that got zero follow-up six months ago
- The magic: it knows everything your Salesforce knows — past interactions, company info, previous event attendance, every note in the record
Delphi (Digital Jason):
- 139,000+ advisory conversations
- Reviews hundreds of VC pitch decks
- Makes product recommendations
- Coaches founders on comp plans and outbound scripts
That’s real. That works. I’m bullish on all of it.
But none of it feels like Replit or Claude. Not yet.
Let Me Explain What I Mean
If you haven’t used Cursor or Replit’s AI agent lately, go do it. Right now. I’ll wait.
These tools don’t just help you code. They understand context. They navigate entire codebases. They reason about what you’re trying to build, not just the line you’re on. They debug across files. They refactor intelligently. They pair program with you in a way that genuinely feels collaborative.
When I use Replit, I feel like I have a digital CTO sitting next to me who has read every file in my repo, remembers everything, and never gets tired. Now this CTO isn’t perfect, and doesn’t check his work as often as I’d like. But we talk about “Replie” in the office as part of our team.
When I use most AI GTM tools, we feel like we have… a really fast ops manager with a template library.
Cursor hit $500M ARR in May 2025 and was at $300M ARR just one month earlier. That’s 60% growth in 30 days. Because it actually works. Because developers can feel the difference.
The best AI GTM tools get 5-7% response rates (which is genuinely good). The best AI coding tools get 30-40% faster feature delivery across entire engineering teams. See the difference in magnitude?
The Honest Truth About AI GTM Today
Let’s be real about what most “AI SDR” and “AI GTM” tools actually are right now:
They’re automation with AI mixed in. It’s hyper automation.
And I don’t mean that as an insult. Automation is valuable! Here’s what they typically do:
- Pull prospect data from various sources
- Use AI to personalize the first line of an email
- Sequence those emails automatically
- Maybe score some leads based on intent signals
- Handle some basic reply classification
- Book meetings when someone says yes
That’s useful. That saves time. We’ve invested in companies doing exactly this — 11x, Artisan, Qualified, and more. They’re growing fast.
But is it an “AI AE”? Is it even an “AI SDR”?
Not really. Not yet.
We went through 47 iterations just to stop our AI from being too aggressive on pricing. It takes 30 days of daily training to nail it. The first 1,000 emails need to be manually reviewed. We still spot-check for 20-30 minutes daily.
Even Agentforce, which has access to all our Salesforce data natively, still requires time. We review draft responses before they go out. We iterate on prompt builder instructions. We think through which campaigns to exclude or target.
This is not “set it and forget it” AI. This is automation that requires constant human oversight.
What Would a Cursor-Level AI GTM Tool Actually Look Like?
This is what I can’t stop thinking about.
Imagine an AI that didn’t just personalize an email, but actually understood your prospect’s business the way Cursor understands a codebase.
It would:
Deeply research and synthesize context. Not just scrape a LinkedIn bio. Actually understand the prospect’s company, their competitive landscape, what challenges they’re likely facing based on their tech stack, hiring patterns, recent news, and market position. The way Cursor understands how all your files connect.
Reason about the entire sales cycle. Not just fire off emails, but understand where this prospect is in their journey. Have they visited pricing? Downloaded content? Been referred by someone? It would think about what touchpoint makes sense right now — the way Cursor thinks about what refactor makes sense given everything else in the code.
Handle complex, multi-threaded conversations. Real deals involve multiple stakeholders with different concerns. A true AI AE would navigate these conversations intelligently. It would know when to bring in technical resources, when to go high and wide, when to focus on the economic buyer. Not just follow a sequence.
Learn and adapt in real-time. The best AEs constantly adjust their approach based on what’s working. A Cursor-level AI GTM tool would do the same. Not just A/B test subject lines, but fundamentally evolve its strategy for a specific account based on signals.
Actually close. Run a demo. Handle objections. Negotiate. Navigate procurement. Do all the things a real AE does that we currently think are too complex for AI.
1mind (from 6sense founder Amanda Kahlow) is attempting this with photorealistic AI avatars that can actually demo and do video calls. HubSpot, LinkedIn, and New Relic are already customers. But even that is early.
Why Aren’t We There Yet?
A few reasons:
GTM is messier than code. Code has syntax. It compiles or it doesn’t. GTM involves human psychology, relationship dynamics, politics, timing, luck. The feedback loops are slower and fuzzier. It’s harder to train against.
The data layer is still immature. Cursor works because your codebase is right there, structured and accessible. The equivalent in GTM — a unified view of all signals about all prospects and all interactions — doesn’t really exist yet. We’re still stitching together intent data, CRM data, engagement data, and enrichment data from a dozen different sources.
The engineering talent went to engineering tools first. This is obvious in retrospect. The best AI engineers wanted to build AI tools for themselves. For their own workflows. Cursor and Replit were built by engineers, for engineers. The GTM AI tools often feel like they were built by GTM people who hired AI engineers — which is subtly different.
We’re still in the “automate the existing process” phase. Most AI GTM tools are designed to do what humans currently do, just faster. But the real unlock comes when AI enables fundamentally new approaches. When we stop thinking about “AI SDR” and start thinking about entirely new ways to create pipeline that weren’t possible before.
The investment is still catching up. AI coding tools have raised billions. Cursor alone is at $29.3 billion valuation. 11x raised $50M, Artisan raised $25M, Qualified raised $163M. Real money, but not Cursor money. Not yet.
But Here’s Why I’m Incredibly Bullish
The gap between where AI GTM is today and where Cursor/Replit are… that’s the opportunity.
Think about it. If Cursor-level intelligence were applied to:
- Account research and ICP identification
- Personalized outreach that actually resonates
- Multi-stakeholder deal navigation
- Real-time coaching for human reps
- Demo delivery and objection handling
- Proposal generation and negotiation
We’re not talking about 20% productivity gains. We’re talking about a fundamental restructuring of how companies go to market.
The AI SDR market is projected to hit $15-47 billion by 2030-2034 depending on which analyst you ask. The AI coding tools market? Already at $3.1 billion in revenue and growing faster than anyone predicted.
The AI AE really is coming. It’s just not here yet.
And when it arrives, it won’t look like “automation with AI sprinkled on top.” It will look like a truly intelligent system that can own a book of business, navigate complex deals, and close revenue.
What This Means If You’re Building or Buying
If you’re building an AI GTM company:
Study Cursor and Replit obsessively. Understand why they feel so much more capable than other AI tools. It’s not just the model — it’s the context window, the agent architecture, the way they maintain state across sessions, the feedback loops that let them improve. Figure out how to bring that same depth of intelligence to GTM.
Cursor has 40-60 employees and is doing $1B+ in revenue. That’s the efficiency you should aspire to.
If you’re buying AI GTM tools:
Buy what’s available today. The automation is genuinely valuable. The time savings are real. Get your team using these tools and learning what works. We spend $500K+ annually on our AI agent stack and it’s worth every penny.
But don’t think you’ve solved AI GTM. You haven’t. No one has. The vendors who tell you they’ve built an “AI AE” are getting ahead of themselves.
The real AI AE — the Cursor of GTM — is still being built. It’s probably being built by someone who hasn’t launched yet. Or by a current vendor who’s about to make a major leap.
Salesforce just acquired Qualified. That’s a signal that the big players see what’s coming.
If you’re a rep:
Learn these tools now. Get comfortable working alongside AI. Because when the Cursor-level GTM tools arrive, the reps who know how to partner with AI will run circles around the ones who don’t.
The best AI doesn’t replace the best reps. It makes them 10x more effective. That’s what we’re seeing with coding. That’s what’s coming for sales.
AI SDRs Can Already Book Six Figure Deals. But That’s Just The Very Start.
I’ve talked to a lot of founders building in this space. The best ones are honest about where they are: “We’ve built really good automation. We’re working toward true AI reasoning.”
That honesty is actually bullish. It means they understand the gap. They’re not confusing “personalized first line” with “genuine intelligence.”
Our AI SDR booked a six-figure deal on a Saturday at 6:02 PM. That’s amazing. A human couldn’t have done that. Agentforce is getting 72% open rates on leads that were completely ghosted — that’s not incremental, that’s transformational.
But could any of them have navigated the negotiation? Handled the procurement process? Upsold when it sensed an opportunity? Not yet.
We’ll look back on 2024-2025 as the “automation era” of AI GTM. The templates, the sequences, the intent signals, the spray-and-pray-but-smarter.
The real AI GTM era — the one that changes everything — is coming in 2026 and beyond. When these tools can actually think. When they can actually sell. When they operate at Cursor level.
That’s when we’ll see entire GTM teams restructured. That’s when we’ll see AI carrying a quota — and hitting it. That’s when the world actually changes.
The AI AE is coming. And it will be incredible.
It’s just not here yet.
