What AI-Powered Sales Hacks Are Your AEs Really Trying to Pull Off With ChatGPT?

We know AEs (and pretty much everyone on the sales team) are already using ChatGPT. DealMentor actually got its start as a GPT we built for one of our own sales teams. The real question is whether you know what your team is using it for, and whether it's actually helping close deals or just another way for them to spin their wheels.
We scoured sales chat rooms across the internet, and here's what we found: AEs and other sales professionals are trying to do real work with AI, because nothing on the market today gives them what they need out of the box.
The use cases that kept showing up
Trigger event research. Salespeople are uploading territory lists into custom GPTs and asking them to search the web daily for mergers, funding rounds, leadership changes, and industry news tied to their accounts. The payoff: opening lines that land because the timing is real, not guessed at.
Call and transcript analysis. Recording calls, dropping transcripts into ChatGPT, and asking it to pull out pain points, buying signals, objections, and next steps. Several people described using this to auto-draft CRM notes and follow-up emails straight from the transcript.
Account and territory research. Public sector and enterprise sellers described feeding board minutes, budgets, and strategic plans into AI and working through a discovery-style Q&A before ever getting a prospect on the phone.
Competitive positioning. A recurring prompt pattern: describe the competitor a prospect just mentioned, ask for a side-by-side comparison, and get a battle card back in under a minute.
Writing and editing. Emails, quotes, internal Slack messages turned professional, QBR content, proposals. The most common line in the whole thread was some version of "if you're not using it for writing, you're wasting time."
Role-play and coaching. A few people built custom GPTs trained on their company's sales methodology and competitive intel, then used them to debrief after real calls. The reps who did this reported meaningfully higher close rates.
Everyone is trying to build their own AI agent to do the work they don't want to do
The more interesting signal isn't the use cases. It's the workarounds people had to build to get there.
Almost nobody is just prompting ChatGPT and getting a good answer on the first try. They're writing custom instructions, uploading PDFs of account lists, re-explaining their ICP every session, and manually copying transcripts between tools. One commenter put it well: building a real custom GPT isn't prompting, it's "encoding its DNA."
And the concerns showed up just as often as the wins:
- Data privacy. More than one person flagged that pasting client data into a personal ChatGPT account doesn't protect the company, even if the seat is paid for personally. If a company hasn't approved a tool, that's a compliance risk regardless of who's paying for it.
- No persistent memory. Every session starts from zero. Territory lists, ICP definitions, and competitive intel all have to be re-uploaded or re-explained.
- Hallucination risk. Several people specifically called out needing a tool with live citations because generic models will confidently make things up.
- It agrees with you. More than one comment noted that AI tools default to positive, encouraging feedback on whatever you give them, which is exactly the opposite of what you need when the real problem is your own blind spot.
That last point matters more than it looks. An AE literally described building a "Virtual VP of Sales" GPT to challenge their thinking and catch blind spots, then admitted they weren't getting much value out of it. A generic model that's trained to be agreeable isn't built to tell a rep their deal is dead or their ICP is wrong.
AI-native solution, or just another way to avoid the real work?
Here's the thing: it's not that AEs are trying to outsource their thinking to AI. They can write their own outbound emails. They can do their own research on a lead. Most of them are good at their jobs. But they're busy, and they want to do the work right the first time, not keep learning the hard way from mistakes that cost them a deal.
That's the real gap. Salespeople don't need AI to think for them. They need a system that already knows their product, their buyers, and their competitors, so they're not rebuilding the same custom GPT from scratch every quarter and hoping it doesn't hallucinate a competitor's pricing.
That's what DealMentor is built to be. Research a prospect and it comes back with a live-web brief and sources, not a guess. Log a call and it drops structured notes straight into the deal, no copy-pasting a transcript between five tools. Hit a competitor mid-call and Handle a Competitor produces the battle card before you're off the phone. And because DealMentor is built to pressure-test your thinking rather than agree with it, it tells you when your deal read is wrong instead of cheering you on toward a loss.
It's the Virtual CRO that AE in the original thread was trying to build from scratch with a custom GPT, minus the re-uploads, the re-explaining, and the compliance risk of pasting client data into a personal AI account.
Build My Sales Program Free No credit card required. Set up your complete sales program in 15 minutes.
The Sales Mentor Always At Your Side
Get 24/7 help to keep your deals moving forward, plus monthly meetings with a live CRO.
Try DealMentor Free