Your Reps Have More AI Than Ever. So Why Are Early Conversations Getting Worse?
By Saagar Sinha, Head of Customer Success & Delivery, Cuvama By Saagar Sinha, Head of Customer Success & Delivery, Cuvama
ISSUE: The Credibility Gap Is Getting Wider, Not Narrower
“We’ve given our teams more AI tools than ever. And somehow, the quality of early-stage conversations has gotten worse.”
Tom Canning, former CCO, Zellis – Panelist, Cuvama London Breakfast Event, 2026
He wasn’t venting or complaining about his team, he was diagnosing a real problem that’s spreading across B2B sales. In a room full of Chief Revenue Officers, Heads of Value, and senior enterprise Account Executives, nobody pushed back or disagreed with him. A few we’re nodding.
That moment stayed with me because it captures something the AI hype cycle consistently skips over, which is that more tools doesn’t automatically mean better selling. In fact, when you add AI without the right foundation, it can actually make your sales conversations worse instead of better.
Over the past two years, Cuvama has run a series of Revenue Leader Breakfast events in New York and London where more than 150 B2B sales and marketing leaders, from frontline Account Executives to VP Sales to Chief Revenue Officers, sat in the same rooms and worked through the same critical question: is AI actually improving early-stage sales conversations, or are we fooling ourselves?
The answer was more complicated than a simple yes or no, but the pattern that emerged was impossible to ignore. AI is amplifying whatever already exists in your sales process, which means if your reps had a sharp, number-backed point of view before AI came along, AI helps them get there faster and more efficiently. But if they were winging it before and relying on generic pitches, AI just gives them more confident-sounding ways to wing it, which means the output improves in volume and polish while the underlying thinking doesn’t actually get any better.
One Chief Revenue Officer at our London event described watching this play out in real time on call recordings, and he said buyers could tell something was off. He heard questions that didn’t sound like anything a human would naturally ask, sellers who were clearly reading from AI-generated prompts, and buyers who started disengaging partway through calls. The worst part was that sellers didn’t understand why they were losing the room because from their perspective, they were more “prepared” than ever before. Buyers can tell when sellers are using AI without thinking, and the credibility gap that creates kills deals before they even start.
IMPACT: This Isn't Just Poor Execution. It's Actively Damaging Your Pipeline.
The problem isn’t theoretical, and it’s not limited to a few bad calls. In value selling, where the entire sales approach depends on earning trust through a specific, number-backed point of view, a credibility gap doesn’t just slow deals down, it kills them at the earliest and most critical stage.
Large language models (the AI systems behind tools like ChatGPT) are genuinely good at producing narratives that sound compelling, structured, and even persuasive. What they are not yet reliable at is producing the specific, quantified business hypotheses that sit behind a credible conversation with a CFO or VP of Operations. Apple’s research on language model reasoning found that these AI systems replicate patterns from their training data rather than performing genuine numerical reasoning, which is a fancy way of saying they’re pattern-matching instead of actually doing math. Finance-focused studies have found error rates above 40% when AI systems try to answer questions involving specific numbers without access to verified data.
Our breakfast panelists confirmed this directly when one said, “Where AI really falls down is trying to estimate or come up with the right numbers.” This matters enormously because in value selling, the numbers are where credibility lives or dies in front of an executive buyer. A compelling narrative that falls apart under scrutiny is actually worse than no narrative at all because it signals that your rep either didn’t do their homework, or they trusted a tool that shouldn’t have been trusted with something this important.
When your reps walk into a room with AI-generated value hypotheses that don’t hold up under questioning, you’re not just losing that deal. You’re teaching buyers that your team doesn’t understand their business well enough to be trusted with strategic decisions, and that perception spreads faster than any win you might recover later. The cost isn’t just one lost opportunity, it’s the systematic erosion of trust across your entire pipeline before your reps even get to discovery.
IMPORTANCE: The Foundation You Build Now Determines Whether AI Helps or Hurts
The solution isn’t to stop using AI or go back to manual research for every deal, it’s to give AI the expert-defined structure and guardrails it needs to produce narrative and numbers together, reliably, from the very first conversation with a prospect. This is where the leaders who are winning have made a fundamentally different choice than the ones who are struggling.
Across every breakfast conversation, every structured interview, and every panel debate we ran, three specific priorities surfaced with enough consistency that we’re confident they represent something real and actionable, not just theory. We documented all three principles in complete detail in the whitepaper, including exactly how leading teams are implementing them and what results they’re seeing. Here’s what I can say without giving away the full framework: they’re not complicated or expensive to implement, none of them require buying new technology or overhauling your entire tech stack, but all of them require doing the foundational work that AI cannot shortcut for you.
The teams getting this right and seeing measurable improvements in early-stage conversion share one thing in common, which is that they’ve invested time in defining what a credible Upfront Value point of view actually looks like for their specific sales motion before they ask AI to help them produce one at scale. They’ve encoded their expertise into the guardrails and structure, not left AI to improvise against a blank page and hope for the best. 73% of the revenue leaders we surveyed said the expert-led approach (combining narrative and numbers as a governed hypothesis) would be significantly influential on their sales motion going forward. The remaining 27% didn’t disagree with the idea or push back on the concept, they were just calibrating where it applies and which deal types benefit most.
The instinct to add more AI tools to your stack is completely understandable, especially when the pressure to show progress on AI adoption is coming from every direction. But the leaders who came out of our breakfasts most energized and ready to take action weren’t the ones who’d added the most tools or spent the most on new technology, they were the ones who’d gotten crystal clear on what a good first conversation actually looks like and were starting to think systematically about how to make it repeatable.
“This is exactly what I try to do myself in all my deals. But I spend hours of research and get to maybe a third of the quality.”
Travis Scott, VP RevOps, Evotix, London Breakfast Event, 2025
The question right now isn’t whether AI can help your reps show up with a point of view, because it obviously can. The question is whether you’ve built the expert structure and guardrails that make AI’s output worth trusting when your reps walk into a room with a CFO or VP who’s going to challenge every assumption. That foundation determines whether your AI investment pays off or silently sabotages your credibility with buyers, and you need to get it right before you scale AI across your team.
The full research, including the three principles that are changing how leading teams approach this, the specific areas of caution that the skeptics raised, and the field evidence on coverage rates and win rates from teams already implementing this approach, is all in the whitepaper. If you’re responsible for a value selling motion and you’re adding AI tools to your process, this research will show you exactly what’s working and what’s quietly undermining your credibility with buyers right now.
This new whitepaper is an essential missing piece for someone who’s actively adding AI to their process; it’s critical intelligence for protecting your pipeline.
Read the full whitepaper: Narrative and Numbers →
Based on primary research from 150+ revenue leaders across Cuvama’s NYC and London Breakfast Series, 2024–2026.
