Your sales team is already using AI. Nobody trained them to
Written byDan Bernoske, Bill Piacitelli, George de los Reyes, Ralph Vetsch and Chris McKenna

Executive summary
"We're seeing a lot of sales and marketing people basically outsourcing their critical thinking to ChatGPT." Dan Bernoske put that to a full table of private equity operating partners, deal partners and commercial leaders at Foley & Lardner in Boston. The first answer came from a fractional CRO: his reps had already bought their own AI, and the company hadn't noticed.
The room's fix was one approved AI tool, one use case proven before the rest, and training on both. Chris McKenna of Foley & Lardner covered what a company still owns once its data goes into a model. Ralph Vetsch, Cortado's sommelier, poured a blind flight of five Cabernet-based wines.
What the room kept asking
What are your reps doing with AI when nobody is looking?
Building their own. Dan described what he sees inside portfolio companies. Reps are coin operated, and they don't want a Big Brother version of ChatGPT. So many set up their own and feed it client data for buyer personas, ICPs and call recordings.
A fractional CRO at a high-tech SaaS company had found the same thing: reps on $20-a-month accounts. His fix started with an audit, and it ended with one tool the company owns.
The fractional CRO's order
Where do you start?
With one use case you can prove. "You've got to treat it like a product launch," Dan said. His own firm struggled with adoption until it narrowed to one or two use cases that started with the user. Once those worked, people trusted the next ones. About a dozen now run inside the firewall.
The fractional CRO's first win was smaller. Recorded calls, with consent, now become a follow-up email with next steps. He called the change "a game changer."
The follow-up email, before and after
Can AI make you better at hiring and coaching?
Only with a person in the loop. An executive recruiter in the room feeds what's known about a candidate into AI to find the weak points. The interview then goes to pressure-testing them. The recruiter says it has even caught their own blind spots.
Another attendee uses the same prep for sales hires, because salespeople can sell themselves better than they perform. Chris McKenna of Foley & Lardner added a warning from his own firm: supervisors have to be in the process more than they think.
Senior people are better at prompting and talking to AI than junior people, full stop.
Where the room split: shut AI off for everyone, or hand reps a tool they're allowed to use?
Shut it off
One attendee's company ran an experiment, then shut off AI access for everyone. Another described firms at the far end, where the rule is no AI at all.
Give them one inside your walls
"If you shut down the access, it's like telling your kid that you can't drink anymore. Well, they're going to go do it someplace else," Dan said. Chris agreed you "can't foolproof it." His firm uses policy, education and IT tracking together, and treats AI as a gas and brake system.
Left open: how much freedom keeps reps from going around you? Dan put it to the room as a question. How do you keep the rules from being so strict that people just go somewhere else?
What you own once it goes into the model
Chris McKenna opened the evening on intellectual property. Three of his points came back during the AI discussion.
Data is the new form of IP currency.
Consumer AI is not confidential
"If you're using the consumer side of AI, not necessarily confidential," he said. A trade secret pasted into an open tool is hard to call a trade secret afterward. The protection comes with the enterprise edition.
Contracts come before patents
A company with no patents or registered trademarks still has IP, and contracts are "the number one that's always in play, every single company, every single time."
Data rights are now a diligence question
In diligence he now asks whether the company's contracts give it the right to use customer data to improve its products. If customers can take their data when they leave, a model trained on it is exposed.
A $300M data and analytics company turned reactive order-takers into enterprise deal closers
$1.6M+ in revenue won, and half of the coached deals advanced.
Read the case studyFour questions to bring to your next portfolio review
These show whether the plan for AI in the sales team can work.
Ask the management team
Which AI tools are our reps paying for themselves, and what client data is in them?
What is the one tool everyone is allowed to use, and who trained the team on it?
Which one or two use cases do we prove first, and how will we know they worked?
Do our customer contracts let us use their data to improve the product?
Who was in the room
Operating and deal partners from Boston private equity firms, commercial leaders and Foley & Lardner's attorneys, at one long table in Foley's offices at the Prudential Center. Ralph Vetsch, Cortado's sommelier, poured five Cabernet-based wines blind. His point for the evening: once you know the structure, you can read what's in the glass. The reveal included a 20-year-old Grand Cru Margaux he had decanted for two hours, and a Caymus he warned most people wouldn't pick blind.
If your reps are paying for their own AI, the company is already behind its own sales team. The question the room left on the table is how much freedom keeps them from going around you.
Ideas in this recap are credited to roles. Attendee names are withheld.
Sign up for our next in-person event
Small rooms of operating partners and commercial leaders. Pick the cities you'd come to, and you'll hear first when a date is set.