Getting Your Reps to Perform at the Ultimate Level? This ISO is Using AI to Train Them

February 2, 2026 | By: Sean Murray

“I‘m unable to get 80, 90 guys to work in the area that I am in so I have to actually max out the team that I have,” says Steven Edisis, CEO of Dynamic Capital. “So my whole thing is onboarding guys and getting them to perform at the ultimate level.”

Edisis says that training small business finance sales reps takes extreme discipline, hours and hours of manual training. Training in the morning. Training in the afternoon. Training at night. Training and then some more training. Some of that training involves roleplaying. Other times it’s live calls. In either case, he’s found there are weaknesses in those systems. For instance, in a roleplay, the trainer has to contend with maintaining rapport with the individual they’re training. Persistent unfavorable feedback, even if warranted, could actually be demoralizing and create tension in the relationship.

“When you’re roleplaying with your buddy, your buddy’s a human,” Edisis says. “After two or three or four times of them getting it wrong, do you just stop correcting it? And you let it go.”

The end result is that they’re not actually in top shape and ready for calls, but they could be deceived into thinking that they are. Meanwhile, live-call training presents another dilemma. If you give the trainee good leads and they mess up, are the good leads wasted? And if you give them really cheap leads and they don’t get anywhere with them, how are they supposed to be judged or learn from it?

For Dynamic Capital, the solution to all of the above has been AI. About a year ago, Edisis began using an AI agent called Ava, a product by Reech AI that’s led by CEO Liran Weissenberg. Ava does for Dynamic Capital what Edisis did for a long time, trains, but on steroids. Trainees, experienced reps, and even veteran pros at the game can roleplay with Ava, a voice AI, in any setup of circumstances they choose. It can be a straight-up cold call, a warm lead, or a follow-up, for example. It can be tuned to easy mode, regular mode, hard mode, or even impossible mode. The best part is that Ava knows how to play the role of a merchant, speaks in real time, and speaks with human-believable tones and emotions. Voice AI technology, once considered clunky or plagued by latency in years past, has finally become virtually indiscernible from a real person.

In a live demo performed for deBanked to show it in action, the AI answered the phone and Edisis went right into the normal flow of business.

“This is Steven with Dynamic Capital. Last week you went online, requested some information for some working capital for your business, how are you today?”

“I’m ok, just a little busy right now,” Ava responded somewhat suspiciously. “yeah, I remember poking around online. I put in my info but I never really got a clear idea what you guys actually do.”

From there, Edisis played it out to a conclusion. When it was over, Ava rated him on his performance, gave him a score, and shared what he did well, as well as things he could have done better. It seemed to understand the relevance of open balances a merchant might have with loans or MCAs, which is key to making it impactful. Ava’s an AI, so a trainee would not be able to attribute the constructive criticism it offered to being singled out or picked on. For instance, if the AI said the trainee needed to work on tone and pacing of speech, there’s no way for the trainee to attribute that to personal bias from a trainer.

“The cool thing is that we can model the AI to behave in very specific scenarios and to have very specific analysis,” said Weissenberg of Reech AI, who created Ava.

Meanwhile, no real leads were wasted in the process. This is especially valuable since Edisis says that he teaches a specific technique—or rather, an art—called the interruption, a very delicate tactic used to keep a call on track. But it only works if delivered correctly, because it involves literally interrupting the prospect while they’re speaking. Learning how to interrupt in the circumstances that call for it is a massive gamble that could not only lead to lost sales revenue but also negative customer feedback if executed poorly. This is a perfect example of where AI training comes into play.

“It’s really nice to [practice it] on Ava versus blowing up and getting negative reviews from a live person,” Edisis said. “So that’s a really cool way that I can train people.”

Weissenberg said that when Ava is being used across a whole sales floor, a sales manager can view transcripts of all the calls, along with feedback and analytics. One could literally have a full-on simulated call center where all the prospects are AI agents being used to train reps.

“The coolest thing about the app is the analytics,” Weissenberg said. “It’s only going to get better too because AI is getting smarter and it’s getting more human.”

“On top of [new trainees], my other guys over the years—some have been here up to ten years—every once in a while they get a little rusty, revert back to bad habits, etc.,” Edisis said. “This allows me to keep them honest. And when some new person comes to me and tells me my leads are bad, I say, “let’s go to Ava… you tell me your pitch, I’ll do the same pitch.”

If Edisis significantly outscores them, it becomes evident that lead quality isn’t the issue, but rather all the other factors that go into having a successful call.

“It keeps some honesty between the program and reality,” he said.