I Called AI Sales Coaching a Dumpster Fire. Then I Built One.

By Tyler Ebnet, Founder and Principal Consultant at revenueify. Co-written, against every instinct I own, with Claude, an AI. Yes, I hear it. Keep reading, the irony is the whole point.

For about 18 months, if you brought up AI sales coaching in a meeting with me, I would have given you the look. You know the look. The one your dad gives the self checkout machine. I sell sales training and coaching for a living, real humans coaching real humans, and I was convinced that handing that job to a chatbot was one of the dumbest ideas our industry had ever tried to sell itself. I said so. Out loud. In rooms with important people in them. Then I went and built one. This is the story of how a card carrying AI pessimist ended up shipping an AI sales coach, and what finally got me to shut up.

First, Let Me Defend the Old Me. Most of It Really Was a Dumpster Fire.

I was not wrong to be skeptical. Look around. “Slop” was literally Merriam-Webster’s word of the year for 2025, and it was a direct nod to the flood of low effort machine generated junk. Mentions of “AI slop” jumped ninefold in a single year, and by some counts more than half the content on the web is now machine generated. We have all gotten that email. The one that opens “In today’s fast paced business landscape” and you can feel the human absence in it, like walking into a room somebody left thirty seconds ago. You can always tell. And here is the uncomfortable part for sellers: your buyers can always tell too, and they are getting sharper at it while we pump out more of it than ever.

And since we are talking about tells, let me get on my soapbox for one sentence. The em dash. AI adores the em dash the way a corgi adores a tennis ball. It scatters them everywhere, these confident little dashes stitching half thoughts together so nothing ever has to fully commit. I have banned them from everything my company publishes. It is a whole thing around here. My team makes fun of me for it. I regret nothing.

So when the AI coaching pitches started rolling in, what I heard was this: let me take the one part of our work that actually feels human, the coaching, and replace it with the slop machine. Most of what I saw earned the eye roll. Generic role plays that sounded like a customer service bot doing improv. Feedback so vague it could apply to literally any salesperson alive. A “methodology” that was clearly seven other methodologies stitched together in a trench coat. It felt like Clippy with a commission plan.

There was also a very practical problem sitting underneath the philosophical one. It is already hard enough to get a salesperson to finish one e learning module. Now I am going to hand them another app to log into? We are obsessive about delivering a world class experience, and I flat out did not believe a robot could deliver that. Hard pass. Or so I thought for about a year and a half.

So Why Did I Stop Plugging My Ears?

People I trust kept pushing me to actually look. Not to buy anything. Just to look. So we put a toe in the water. Then another toe. We started weaving AI into our existing training, because it turns out AI is genuinely spectacular at the boring, time sucking prep work that eats a rep’s morning. We liked it so much we built a whole hands on bootcamp around it called Prep Like a Pro. Using AI to prep is a no brainer. Using AI to coach still felt like a bridge too far.

But the more we worked with our curriculum designers and content people, the more I started to see a version that was not garbage. Not a bolt on. Not a rented model with our logo slapped on it. Something we could actually control, end to end. So in March, fairly quietly, we launched it. And I want to be honest about my emotional state at the time. I was deeply, professionally suspicious of my own product. That is a weird way to ship something. It is also, in hindsight, exactly why it turned out okay.

The Three Things That Turned a Hardened AI Pessimist

Three things flipped me. Not a slick demo. Not a fake buyer persona named Brenda from Procurement. Three real things, in order.

One. Methodology first. The reason most AI coaching is slop is that it has no spine. It scrapes a little of this framework, a little of that one, whatever ranked on Google that week, and blends it into a Frankenstein that contradicts itself by paragraph three. Ours is built entirely on Customer Focused Selling®. One method. The exact same one your team is already being taught in the room. So the AI is not out there freelancing its own opinions about selling. It is reinforcing the plays your human coaches actually teach. That was the first time my shoulders came down from around my ears.

Two. Reps do not need a pep rally. They need reps. We have never been a one time, rah rah, trust falls and a branded water bottle kind of shop. Anybody can get a room fired up for an afternoon. Then Monday happens, and the fire is out, because reps forget about 70% of training within a week and 87% within a month. The magic is never day one. It is month two and month three, when a rep gets a small win and you are right there to reinforce it and build on it. That is the entire argument for why coaching beats training. The catch is that you cannot be right there at 7pm on a Tuesday when a rep finally has ten minutes to practice. An AI coach can. And reps actually like it, because it is self serve and nobody is watching them fumble the first three attempts.

Three. The part that actually got me. The data. Here is the plot twist I did not see coming. The AI coaching helps the sellers, sure. But the real gift is what it quietly hands the sales leader. For years, the honest answer to “where does my team actually need help” was a shrug and a gut feel, mostly because managers spend under 5% of their time coaching, roughly 36 minutes a week per rep. We already diagnose this up front with our A.I.M. Assessment. But now I could see it live, all the way through a program, not just at the start.

I will not name the client, but we are running this right now with a strong field sales team, about eleven reps. Inside the first month, the platform surfaced something a busy manager would never have had the time to piece together across a stack of one on ones: the whole team shared the same small habit early in discovery, a spot where slowing down just a beat before moving ahead would open the conversation up even more. Honestly, that is the kind of pattern you hope to find. It is not eleven separate things to chase one at a time. It is one focused group session on a Thursday. Do you have any idea how much time that single insight saves? The tool showed us the flip side too, the places reps were genuinely excelling, so we knew exactly who should be teaching everybody else. It was not a report card. It was a map. That is the moment the pessimist quietly tapped out.

What to Actually Look For in an AI Sales Coach (Most of Them Flunk This List)

Okay. So I came around. That does not mean the market did. Most of it is still a dumpster fire, and I say that as someone who now sells one of these. If you are going to try an AI sales coach, and after 18 months of resisting I am telling you that you probably should, here is the list I wish somebody had handed me before I wasted a single meeting.

It has to be grounded in a method. Not “a coaching philosophy.” A named, repeatable methodology your team already runs. Without one, the AI has nothing real to grade against, so it defaults to “great job, sound more confident” nonsense. As Dave Kurlan put it, AI is great at “or” and terrible at “and,” and coaching only works when it is timely, relevant, and actionable. If the tool is not coaching your method, it is quietly coaching against it.

The simulations have to be built on your world. This is the big one. Generic role plays do not connect, because your reps know they are fake and they check out. We build simulations from our customers’ actual deals and actual situations. Real example: we work with a landscaping company that kept stalling out because one specific tree was out of stock. So we did not hand their reps a generic “handle the objection” drill. We built a simulation, using our six step process for difficult conversations, that speaks fluent landscaper and argues about that exact tree. Their reps got to practice the conversation they were actually about to walk into. That is the whole game.

Instant feedback, plus a manager view. The rep needs to know how they did the second they finish, not next quarter. And the manager needs the dashboard, because that is where your return on investment actually lives. Practice that nobody measures is just a very expensive video game.

The simulations have to get updated. Nobody wants to run the same scenario forty times. Reps get bored, they start gaming it, and it goes stale like every app you downloaded and opened exactly twice. Your business moves. The practice has to move with it. If the scenarios are frozen in time, congratulations, you bought a screensaver.

It has to keep the human in. Ground it in behavior science, or you will train a team of robots to sell to humans, which goes about as well as it sounds. We build ours on DiSC® so reps learn to read and flex to how a real person actually buys, instead of reciting a script at them. The entire point of the AI is to make your people more human in front of a customer, not less.

One bonus tell, since we are already here. Look at how many “tools” your team pays for and never opens. Roughly 67% of software features go completely unused, and reps actively use about three of the ten plus tools you buy them. If your AI coach becomes just another login, it does not matter how brilliant it is. It has to meet reps where they already are, or it dies in a browser tab.

Okay, My Turn to Shill. I Fought This for 18 Months. I Earned Thirty Seconds.

If you made it this far, you have earned the sales pitch, and honestly so have I. Here is what we built and how it maps to that exact list, because I would be a hypocrite otherwise.

We update the simulations every month, based on what you and your team actually want to work on next. The coaching itself is built in three layers. Layer one is Customer Focused Selling®, the method, so everyone is grounded in the same playbook. Layer two is your world: your customers, your buyer profiles, the situations your reps are genuinely walking into, and whatever your company is focused on this quarter. Layer three is where it gets good. We build a connected experience that compounds on itself. Most of our customers run it about twenty minutes a week, four focus areas that build on each other, and then the month resets and we go again. It is a rhythm, not an event.

You get deep manager feedback and individual feedback. And here is the piece I flat out insisted on, the part most platforms will never bring up because it costs them money: a real human from revenueify still reviews the interactions. The AI is phenomenal at volume and consistency. It is not a replacement for a person who can spot the thing the model glossed over. We want to see where your people need more help, so a human still looks. That is not a limitation of the technology. That is the point of it.

We still build in the community piece, because salespeople are tribal and the best coaching a rep ever gets is another rep saying “here is what is actually working for me right now.” And we will put the whole thing wherever your team already lives. Want it inside Salesforce? Fine. Teams? Great. The rule is stupidly simple. Meet reps where they are, or watch them not use it.

Want to See It? No Gimmicks. I Promise.

Here is where a normal blog drops a giant yellow “Start Your Free Trial” button and fires you into a self serve funnel that emails you four times before lunch. I could do that. But that would make me the exact thing I just spent two thousand words dunking on. Generic. Automated. The self checkout of software.

So no. I am not going to just hand you a signup link and call it a relationship. Give me fifteen minutes. Tell me your world, your customers, the deal your reps keep losing on repeat. I will build you a demo that is genuinely relevant to your team, and I will have it ready the very next day so your people can actually test drive the thing. Then you can tell me whether the pessimist was right the first time, or the second time.

Book 15 minutes and I’ll build you a real demo →