I just watched a founder blow their shot at an easy deal in real-time with me.
- The CEO of a marketing platform sent me what was actually a pretty strong pitch. Good positioning, clear value prop, strong social proof. The kind of outreach that usually gets a response.
- And in fact, he offered to solve a real problem we need solved. The timing of the pitch and value prop was perfect.
- Then he wrote: “I know you’re busy as SaaStr Annual is 11 weeks out.”
Dude. SaaStr Annual + AI Summit 2025 already happened. Way back in May.
Game over.
Not because of the mistake itself, but because of what it revealed: his AI-powered sales system wasn’t actually being audited by humans. And in B2B sales, that’s a fatal flaw.
The AI Sales Paradox
Here’s the thing—AI in sales can be absolutely transformative. I’ve seen it work magic:
- Personalization at scale that would be impossible manually
- Lead scoring that surfaces hidden gems
- Response optimization that dramatically improves reply rates
- Follow-up sequences that nurture prospects without human intervention
This CEO’s pitch actually demonstrated several of these strengths. His AI had clearly analyzed my content, understood my audience, and crafted messaging that resonated with my specific situation.
But AI also has a catastrophic failure mode: it confidently states things that are completely wrong.
The 90% Problem
When the CEO replied to my feedback with “you said it yourself, AI is really good 90% of the time,” he missed the point entirely.
90% accuracy sounds great in theory. But in sales, that 10% error rate isn’t distributed evenly. It tends to cluster around the most important, most visible, most easily-verified facts.
And when you’re wrong about something basic that your prospect knows better than anyone else, you don’t just lose that deal—you lose all credibility.
The Audit Imperative
The solution isn’t to abandon AI in sales. It’s to build proper human oversight into your AI-powered sales motion.
Here’s what actually works:
1. Fact-Check the Obvious Stuff
Before any AI-generated outreach goes out, have a human verify:
- Event dates and locations
- Company details and recent news
- Industry-specific terminology
- Competitive landscape claims
This takes 30 seconds per message and prevents 90% of credibility-destroying errors.
2. Test Your AI’s Knowledge Boundaries
Regularly audit your AI system with questions about:
- Recent events in your prospect’s industry
- Current market conditions
- Specific company milestones
- Technical details about their business
When your AI gets something wrong in testing, fix the training data before it embarrasses you in the field.
3. Build Confidence Scoring
Train your AI to flag when it’s uncertain about specific facts. Better to send a slightly less confident message than a confidently wrong one.
4. Create Review Checkpoints
For high-value prospects, always have a human review AI-generated outreach before it goes out. The ROI on this review time is massive when you’re targeting enterprise accounts or key strategic partners.
A deep dive with $1B+ Owner’s CRO Kyle Norton on how they do this here:
The Trust Tax
Every sales interaction is fundamentally about trust. When your AI makes a basic factual error, you’re not just losing that opportunity—you’re paying what I call the “trust tax.”
The prospect now questions everything else you’ve said. Your market positioning, your customer success stories, your product claims—all of it becomes suspect.
And here’s the kicker: the more sophisticated your AI sounds, the higher the trust tax when it screws up. If you’re positioning yourself as an AI-powered solution, getting basic facts wrong is exponentially more damaging.
The Right Way to Use AI in Sales
Don’t get me wrong—AI can be incredibly powerful in sales when used correctly:
Use AI for:
- Initial research and lead qualification
- Draft creation and message optimization
- Response timing and cadence management
- Performance analysis and iteration
Always human-verify:
- Specific dates and events
- Company-specific details
- Industry facts and figures
- Competitive comparisons
The Meta Lesson
This CEO’s mistake wasn’t just about SaaStr Annual. It was about not understanding that in B2B sales, you don’t get a second chance to make a first impression.
When you’re selling to sophisticated buyers, they’re evaluating not just your product, but your operational competence. A basic factual error signals that either:
- You don’t have proper quality control processes
- You’re not paying attention to details
- You’re over-relying on automation without human oversight
None of these are good signals for a potential vendor relationship.
The Path Forward
AI in sales isn’t going away—nor should it. But the companies that win will be those that combine AI’s scalability with human judgment and oversight.
The goal isn’t to eliminate AI from your sales process. It’s to eliminate the scenarios where AI can blow your shot.
Build systems that let AI do what it does best (scale, personalize, optimize) while ensuring humans catch what AI does worst (context, nuance, basic fact-checking).
Because in B2B sales, 90% right is 100% wrong. Especially when you don’t already know and trust the vendor.
