While companies are going all in on AI to boost their sales, hitting revenue targets has never been harder.
To improve how pitches land, enterprises are embracing a tool that has gained traction since ChatGPT launched: AI sales coaches.
These systems simulate customer conversations, letting sales reps practice pitches in real time with AI personas that act like buyers. The tools then score performance and deliver personalized feedback, supplementing manager-led coaching and static training with an interactive, repeatable practice.
The shift comes as a key part of sales training is breaking down. Companies are cutting middle managers to boost efficiency, and those who remain oversee larger teams and spend less time mentoring employees.
That squeeze in coaching leads to lower engagement and higher turnover. A 2025 retention report from the Work Institute, an HR research firm, found that 18% of workers cited insufficient professional development opportunities as the main reason for quitting their jobs. A lack of sales coaching could leave sales teams with less support to improve.
Now, companies are embedding AI sales coaches into training to help fill that gap. These tools can standardize feedback, give reps more chances to practice, and make performance measurable. Early results show faster onboarding times and stronger communication.
But the shift raises questions on how much of sales training can be scaled through simulation — and what gets lost when coaching starts to feel less human.
Practicing the pitch
At ServiceNow, an enterprise AI platform, an AI sales coach is taking over repetitive training work.
Jayney Howson, the company’s senior vice president of global workforce skills and talent readiness, said the tool is built into ServiceNow University, its internal training platform, where roughly 8,000 sellers use it for certifications and product training. Before AI, training relied on webinars, self-guided courses, and manager feedback through a quick “vibe check.” That subjective evaluation method varied widely, slowing the time it takes to prepare sellers for real calls.
Now, sellers can practice role-play simulations as often as needed and receive metrics that score their skills and quantify their improvement.
“High effort learning is very difficult to deliver,” Howson told Business Insider. “It’s brilliant for that.”
In one exercise, a seller interacts with an AI persona named Jordan, a midlevel buyer who asks: “What are you here to talk about?” The sales rep then delivers a pitch, handles follow-ups, and is scored against a rubric. The system flags gaps, like not asking enough questions to understand a customer’s needs, and sellers repeat the exercise until they improve.
Since introducing the tool earlier this year, Howson said about 90% of ServiceNow’s sellers use it. He added that the time it takes employees to get up to speed has dropped from three months to six weeks.
By moving away from manager-led “vibe checks” of sellers’ skills to quantifiable metrics, Jayney said the AI coach has significantly improved confidence in sales readiness. With training automated, managers can focus on higher-value mentorship.
“AI should help you manage so you can lead,” Howson said.
Turning communication into data
At Braintrust, a sales training firm, AI sales coaching is used to break down how reps communicate and turn it into measurable data.
The firm, which trains clients in pharma, SaaS, and finance, uses a tool called Yoodli to conduct and analyze buyer-seller conversations. It tracks pacing, filler words, and conciseness, then maps those behaviors to its internal sales framework.
The system is trained on Braintrust’s proprietary data, including training videos, rubrics, and books on its neuroscience-based sales approach.
Jeff Bittner, Braintrust’s director of digital and AI, told Business Insider that practicing with AI lowers the stakes and allows sellers to “practice and screw up” without performing in front of others.
Yoodli’s AI personas are tailored to specific customers. For one pharmaceutical client, Braintrust created an AI oncologist persona to benchmark performance. Within three months, Bittner said the cancer diagnostics company saw its team’s ability to “personally connect” rise from 10% to 84%.
Bittner also pointed to a security company that landed larger deals after improving “problem quantification,” a skill practiced in the AI role play system that helps sellers calculate the financial cost of a customer’s pain points.
The limits of simulation
Despite early wins, the tools are quick to show their limits as machines. Braintrust found that interacting with AI avatars doesn’t feel like talking to real humans. “There’s just something not right about them,” Bittner said, adding that the AI’s lips “get a little jittery” as it speaks and makes the conversation feel unnatural.
At ServiceNow, Jayney said she is concerned AI could weaken manager relationships if leaders avoid difficult performance conversations. “Managers are leaning out of those conversations,” Jayney said. “That’s a genuine risk.”
There is also a chance of overreliance. Jayney said AI can help sellers practice and measure performance, but it cannot replace the judgment needed to understand a client and adapt in real time.
“You can’t automate curiosity or the wisdom of knowing a conversation is going the wrong way,” she added.
The future of AI sales coaching
Despite risks of eroding the human touch that come with mentorship, both companies are pushing toward deeper integration.
ServiceNow is working to tailor training to specific regions and customer contexts. Braintrust is analyzing real sales calls alongside AI simulations to see whether strong performance translates into revenue.
While the tools show potential to standardize practice and measure performance, they can’t fully replicate the human guidance needed to train strong sellers in real conversations.
Companies, however, don’t see it as a tradeoff.
“It’s not about replacing their job,” Bittner said. “It’s about assisting them.”
Read the full article here
