When Good AI Goes Bad
AI has quickly become embedded in the modern B2B sales tech stack – powering prospect research, call summaries, outbound sequences, and even business case creation. It promises speed, scale, and smart insights. But the reality? It often falls short where it matters most.
At Cuvama’s NYC breakfast event, we gathered over 30 sales and value leaders to discuss how AI is being used today. What emerged wasn’t just success stories – it was a candid look at the cracks in how AI is deployed.
We heard frustration. Missed opportunities. And real-world examples of when automation didn’t just underdeliver – it actually damaged trust, deals, and brand reputation.
This blog captures the war stories that surfaced – and more importantly, the lessons sales leaders need to learn if AI is going to be a true enabler of value-led selling.
Let’s start with where things go wrong.
The Curse of Superficial Personalization
AI-generated emails often look the part – well-structured, polite, even timely. But beneath the formatting and tone, many fall short where it matters most: relevance. That’s because most AI tools don’t truly understand your buyer’s world. They scrape public data, latch onto industry jargon, and plug in a few familiar keywords. The result? Messages that sound personal on the surface but miss the mark entirely.
Here’s where it breaks down:
No real grasp of the account’s priorities or strategic context
Reliance on surface-level industry themes
Zero integration with your internal knowledge – customer stories, product positioning, segment challenges
Just because it uses your name and company doesn’t mean it understands your pain.
‘Only 15% of B2B buyers say vendor outreach feels genuinely personalised.’ – Gartner Future of Sales 2025 Report
To earn attention, let alone trust, messages need to reflect real insight. AI can help scale personalisation, but only if it has access to the context that makes personalisation meaningful.
Over-Automation = Lost Empathy
“It sounded robotic. Like a machine was just checking boxes.”
AI has made it easier than ever to automate follow-ups, streamline sales cadences, and even generate discovery questions. But when automation goes unchecked, it does more harm than good. What begins as a time-saving tool can quickly become a relationship killer. Buyers can tell when your outreach lacks intent. Whether it’s a cold LinkedIn message that reads like a template or a discovery call that follows a rigid script, over-automation strips away the emotional intelligence that makes great selling… well, human.
And the data backs this up:
‘64% of B2B buyers say overly automated communication is a top reason they disengage from a vendor.’ – Forrester 2023 Buyer Experience Survey
Where it shows up:
- Bland, templated LinkedIn DMs
- Scripted call openings with no space for curiosity
- Auto-generated follow-ups that feel hollow
If sales is about building trust, automation must be wielded carefully. The more we scale our outreach with AI, the more important it becomes to leave room for real connection, thoughtful improvisation, and emotional nuance.
Chasing Volume, Not Value
“It helped me send 100 messages. I got blocked by 30 of them.”
AI has made it effortless to scale outbound activity – but volume alone doesn’t win deals. In the rush to fill the top of the funnel, many teams fall into the trap of prioritizing speed over strategy. The result? More messages, fewer conversations. Relevance suffers, signals get missed, and buyers tune out. Worse still, brand trust begins to erode as prospects are hit with impersonal, ill-timed outreach.
This pattern is becoming all too common:
- Spray-and-pray outbound tactics
- Shallow targeting logic
- Weak conversion to real conversations
As one sales leader put it, “We’re scaling activity, not outcomes.”
And the data confirms it:
‘Outreach volume has doubled with AI, but reply rates are down 28% on average.’ – Salesloft 2024 State of AI in Sales Report
AI isn’t a cheat code for pipeline. It’s a tool for precision. To drive meaningful engagement, every touchpoint needs to be considered, contextual, and rooted in real insight – not just automated output.
Insights Without Context
“It gave me stats. But not a story.”
AI can surface an ocean of information – company news, industry benchmarks, usage patterns. But here’s the problem: data alone doesn’t move a deal forward. What actually resonates with buyers is a story – one that clearly connects their specific pain to a credible outcome. That’s where most AI tools fall short. They generate generic insights, but without the internal context of your value proposition, customer proof, or competitive differentiators, those insights lack impact.
In other words, if your AI doesn’t know what you know, it can’t tell your buyer what they need to hear.
‘62% of sales reps say their AI tools lack access to internal enablement content.’ – LinkedIn State of Sales 2024
That missing link turns what could be a powerful message into something forgettable. Because the best AI isn’t just informative – it’s contextual. It learns from your actual sales conversations, reflects your internal positioning, and evolves the story as the deal unfolds.
Until that happens, sellers are left stitching together the narrative on their own.
Real AI Horror Stories: When Automation Backfires
The war stories we heard in New York weren’t about broken software or faulty algorithms. They were about misapplied tools – and the real-world consequences of using AI without context, curation, or common sense.
Here are just a few examples that stood out:
- A rep sent a “personalised” email congratulating a prospect on their competitor’s funding round – the AI had scraped the wrong name from a news feed.
- A buyer received the exact same intro message from three different SDRs at the same company in the same week – all triggered by a cadence tool running without oversight.
- A business case, built using a GPT-based tool, included impressive ROI metrics… for the wrong industry. The champion had to rewrite the entire thing before presenting it internally.
Each of these moments chipped away at credibility. They wasted time. And more importantly, they made it harder for those sellers to have the real, value-led conversations that win deals. These aren’t rare outliers. They’re symptoms of a broader issue: AI tools acting without insight, and sellers trusting the output too much.
So What Can We Learn?
The stories shared at our NYC event weren’t about technical malfunctions – they were about trust lost, context missed, and strategy abandoned. If there’s one takeaway from the discussion, it’s this: AI isn’t inherently good or bad – it’s a mirror. It reflects the quality of the inputs, the structure of the systems it supports, and the judgment of the people who use it.
Here’s what the most thoughtful sales leaders are starting to get right:
Don’t Delegate Strategy to AI
AI should accelerate execution – not replace thinking. It can suggest talking points, summarise calls, and even draft proposals, but it doesn’t understand deal nuance, politics, or timing. Top sellers win because they interpret, adapt, and respond in real time. That responsibility can’t be outsourced.
Context Beats Speed
High-output AI without context is just noise at scale. Generic personalization, vague pain points, and misaligned messaging hurt credibility. The best AI systems are connected to internal knowledge – like win stories, segment nuances, and competitive insights – so they can generate messaging that feels like it came from your best rep, not a bot.
Automation Demands Curation
It’s tempting to “set and forget” your AI workflows – but unchecked automation spreads mediocrity fast. A cadence filled with irrelevant steps will scale faster with AI, not better. Leaders must curate libraries of content, review prompt libraries, and actively enable teams to edit and contextualise AI output before it reaches the customer. Human QA isn’t optional – it’s your quality filter.
Buyers Deserve Better
Today’s buyers aren’t just researching – they’re using AI too. That means they’re more informed, more skeptical, and less tolerant of boilerplate outreach. If your AI-assisted message doesn’t add value, they’ll move on. Fast. Your first impression now matters more than ever.
Don’t Blame the Tools. Fix the Strategy
These aren’t one-off horror stories, they’re symptoms of a deeper issue. When AI is applied without context, without oversight, and without strategic intent, it doesn’t accelerate sales – it undermines it.
A joint study by Databox and Jasper.ai found that nearly 50% of users cited “quality and accuracy of outputs” as their top challenge with AI tools, followed closely by “lack of human touch.” – Databox x Jasper.ai Report
If we want AI to actually help sellers sell, we need to redesign the system around how real selling works. What sales teams need isn’t just more automation. They need:
- Context-rich systems that understand the buyer’s world
- Internal knowledge integration that embeds your value stories and proof points
- AI that works with sellers, not instead of them – adapting to each deal, evolving the narrative, and reinforcing the value story from first call to close
That’s the new standard.
