AI in sales has exploded, and so has the hype.
Every week, there’s a new promise that “artificial intelligence will revolutionise your sales process,” but anyone actually selling knows the reality is far more nuanced.
At Cognism, we’ve been hands-on with AI for a while now.
From ChatGPT and Gong AI to our very own CognismGPT, our sales team has experimented with these tools in real-world sales processes, prospecting, analysing sales calls, automating admin, and personalising outreach.
Sometimes, it’s been game-changing. Other times… not so much.
So, instead of a generic AI article, we’re sharing real stories from the Cognism sales team.
Five sales team members reveal where AI has helped them sell smarter, and when it’s missed the mark entirely.
These aren’t theory-based examples or polished success stories.
They’re genuine wins and fails that show what happens when AI meets the fast-paced reality of B2B sales teams, where timing, tone, and human judgement still make all the difference.
When Senior Sales Manager Isa Sher first tried to use AI to score his sales team’s DISCO calls in a sales performance tool, the results looked promising on paper.
In practice? Not so much.
AI was supposed to help Isa identify which reps were nailing discovery conversations and which needed coaching. But the scores weren’t as accurate as he’d hoped, rewarding sales calls that missed key points and overlooking those that nailed qualification.
The problem?
AI couldn’t read nuance, the tone of a curious question, the empathy in a rep’s voice, or how a prospect reacted. It treated sales conversations like data points, not dialogues.
Isa summed it up best:
“AI can surface trends, but it can’t yet understand human connection. That’s where you still need a manager’s ear.”
The takeaway? AI insights are only as valuable for your sales processes as the human interpretation behind them.
We’ve all seen them, the generic, AI-written LinkedIn messages that land in your inbox and immediately get ignored.
For Account Executive Johnny Stiffell, they’re the ultimate example of where AI can go wrong in sales. He said:
“When people use artificial intelligence to send out copy-paste DMs, it shows. There’s no personalisation, no relevance, and definitely no human touch.”
Johnny’s seen messages like:
It’s not just ineffective; it can actually damage a rep’s credibility. AI-generated outreach without context or intent becomes noise instead of nurture.
The takeaway?
AI can’t automate authenticity. In sales, empathy and research still win every time.
For Sales Coach and Why Did It Fail? Podcast host, Shivan Pillay, experimenting with AI has been a year-long learning curve, and one lesson stands out:
Your results are only as good as your prompts.
Early on, Shivan tried to get AI to analyse sales call data from the Cognism sales team at a granular level.
He wanted it to identify specific phrases reps used in successful deals versus those that didn’t close.
But the output he got back was vague summaries, broad themes like “strong discovery questions” or “needs better qualification”, not the verbatim insights he needed.
“I realised I was asking AI the wrong thing. It was never designed to pull exact quotes from call data; it summarised patterns instead. That was on me, not the model.”
It’s an easy mistake for sellers to make: treating AI like a mind reader instead of a tool that needs clear, structured input.
The lesson? If you want precise answers, you need precise prompts to help.
For Enterprise SDR Ellie Childs, AI has become an everyday assistant for researching accounts before outreach. But, as she’s learned, you can’t take everything it says at face value.
Ellie once used ChatGPT to prepare for a call with a high-value prospect, asking it to summarise recent company updates.
The AI produced a polished overview, complete with references to a “new product launch” that, as it turned out, had been discontinued months earlier.
Ellie said:
“The prospect actually corrected me on the call., It wasn’t a huge deal, but it instantly broke the flow of the conversation.”
Moments like that highlight a key flaw: many generative AI tools rely on static or outdated data. In fast-moving industries, that can make your outreach feel disconnected or even careless.
The fix?
Ellie now always double-checks AI outputs against Cognism’s Sales Companion to confirm company updates, org changes, and news in real-time.
Lesson: AI can supercharge your prep, but accuracy still depends on the data behind it.
For Account Executive Stevie Griffiths, the worst AI misfire isn’t hidden in a dashboard; it’s sitting in her LinkedIn inbox. She said:
“I’ve had a flood of AI-generated voice notes from reps lately. The second you hit play, you can tell it’s robotic with no personality.”
“I don’t even finish them anymore. I just click off. It’s the definition of spray and pray.”
As Stevie says, these LinkedIn voice notes aren’t personal at all. The rhythm, pauses, and lack of genuine energy make it clear it was generated, which is an instant turn-off.
Stevie’s advice? Use AI to inform your outreach, not deliver it. A human voice still matters, literally.
While AI fell short in scoring his sales team’s calls, it excelled in helping Isa build something sales leaders struggle with: onboarding materials.
Tasked with creating documentation for new hires, Isa started with a skeleton outline: key objectives, target behaviours, and required competencies.
Isa then used ChatGPT for sales to structure it into comprehensive onboarding workflows, with clear expectations and practical exercises.
What could’ve taken weeks of writing and formatting became a collaborative process. Isa defined the strategy, and AI helped bring it to life.
Example: you can feed ChatGPT a prompt like:
“Create a two-week onboarding plan for new SDRs focused on pipeline generation and discovery call excellence. Include daily activities, key metrics, and a checklist of behaviours.”
The result? Professional, structured documentation you can easily tweak for future hires.
As Isa puts it:
“AI doesn’t replace your experience; it amplifies it. It gave me a head start and freed me to focus on coaching instead of admin.”
Where Johnny really sees AI shine is in admin automation, specifically through CognismGPT, a custom AI tool built in-house by Cognism’s own Jafar Orujov.
Johnny uses it daily to streamline his workflow: summarising Gong transcripts, capturing next steps, and aligning notes with the MEDDICC framework.
He even uses it to generate polished follow-up emails that only need light editing. In Johnny’s own words:
“I use the upgraded ChatGPT version and pay for it myself because it’s that valuable. I can drop in a Gong transcript, ask CognismGPT to pull out next steps or action items, and it saves me hours every week.”
Example prompt Johnny uses:
“Summarise this Gong call transcript using MEDDICC and draft a follow-up email to the prospect referencing the next steps discussed.”
The result? Consistent, on-point communication and more time to focus on what matters most, selling.
It’s a perfect example of how AI sales tools and human expertise can work together, with CognismGPT doing the heavy lifting and Johnny adding the personal touch that turns automation into action.
Once Shivan refined his prompts, the game changed.
He began analysing a full year’s worth of call data, asking AI to identify trends in how Cognism’s AEs and AMs navigated the MEDDICC process, everything from multi-threading and stakeholder mapping to how top performers framed value during sales calls.
Using AI, Shivan could quickly see what top-performing reps were consistently doing well and where underperformers were hitting barriers.
For example, prompts like:
“Analyse these call summaries from top-performing AEs & AMs and identify common themes in how they handle decision criteria and champion development.”
The output gave him actionable patterns, not just metrics, but language, tone, and approach. Shivan said:
“AI helped me quantify what ‘good’ actually looks like. It highlighted the behaviours that drive momentum in deals, so we can coach to replicate them.”
The takeaway?
AI in sales isn’t about replacing analysis; it’s about accelerating it. By letting AI crunch the data, Shivan can focus on translating insights into enablement strategies that move the needle.
When used the right way, AI has transformed how Ellie personalises her outreach.
Instead of stopping at surface-level data, she uses ChatGPT and Research by Cortex AI to interpret company information and turn it into relevant talking points.
For instance, she might prompt ChatGPT:
“Tell me which companies compete directly with Company X and summarise what differentiates them.”
“List any leadership changes or new hires at Company X in the last three months.”
Armed with this context from ChatGPT and Research by Cortex AI, she crafts personalised openers that stand out from the standard noise.
These look like:
“Congrats on expanding your EMEA team. I noticed your new VP of Sales previously scaled an outbound model using intent data. That’s exactly the kind of signal we help sales teams act on faster.”
It’s researched, relevant, and rooted in insight, not fluff. Ellie said:
“AI helps me find the why behind my outreach. It connects the dots faster, so I can focus on building the relationship.”
So, what’s the takeaway?
AI isn’t replacing prospecting skills; it’s refining them. Used smartly, it helps reps go deeper, faster, and with confidence that their outreach is grounded in value.
While Stevie’s sceptical about AI replacing the human side of sales, she’s quick to admit it’s incredibly useful for performance improvement.
Lately, Stevie’s been using Gong’s AI sales assistant to review her calls through the lens of MEDDICC, identifying which parts of the framework are strong and which need more depth.
She told us:
“After each call, I can see where I didn’t dig deep enough into metrics or the decision process. It’s like having an extra layer of quality control.”
By prompting the AI to highlight missing MEDDICC elements, Stevie gets a quick, actionable summary that guides her next conversation. For example:
“Review this call and highlight any MEDDICC criteria that weren’t fully covered.”
The feedback helps Stevie focus her future calls on gathering complete information, driving deals forward with more structure and less guesswork. She said:
“It’s not about letting AI coach you. It’s about using it to spot gaps faster, then doing the real work yourself.”
The takeaway? AI can’t replace real sales coaching, but it can make reps more aware, proactive, and prepared.
In Stevie’s words:
“It’s like having a mirror for your sales process, one that never gets tired of listening and supporting your sales cycle.”
Across every story, one truth stands out: AI doesn’t sell; people do.
From Isa’s onboarding workflows to Johnny’s follow-up automation.
From Shivan’s data analysis to Ellie’s personalised outreach and Stevie’s performance reviews, the wins came when AI worked with the rep, not instead of them.
The misses? They happened when AI tried to take over. When automation replaced authenticity. When the data wasn’t checked, or the message lost its human spark.
The future of sales isn’t about choosing between AI and people. It’s about combining the precision of technology with the empathy of experience.
That’s how sales teams sell smarter and strengthen their sales cycle.