Coaching begins with something you noticed. Several of the things you used to notice have stopped showing up.
Every coaching conversation starts the same way. You observe a symptom, you work out what it points to, and you coach to that. A seller shows up underprepared for a discovery meeting. A proposal covers everything and lands on nothing. A deal stalls after a meeting that felt strong. Each of those is a signal, and each one tells a manager where to aim.
That whole chain depends on the symptom being visible.
Think about what a discovery brief used to tell you. If a seller hadn't really done the work, you could see it. The research was thin, the questions were generic, and there was nothing in there specific to the person they were about to sit across from. That document was doing two jobs at once. It got them ready for the meeting, and it showed you how they think.
A generated brief looks nothing like that. It arrives thorough, well organized and properly formatted, whether or not the seller learned a single thing while it was being made. The document that used to expose a gap now covers it up.
That leaves a manager in an awkward spot. There is no symptom to notice, so nothing gets written down, and nothing gets coached. The seller who never quite learned how to prepare still hasn't learned, and you both find that out in a meeting where it's too late for anyone to help.
You probably won't hear about it from them, either. Just over half of the people who use AI at work, 52%, say they are reluctant to admit it on their most important tasks, and 53% believe that using it makes them look replaceable. Those figures come from the Work Trend Index by Microsoft and LinkedIn, drawn from 31,000 responses across 31 countries1.
Sit with what that combination means in practice. On exactly the work that would benefit most from a coaching conversation, half your team has a good reason to stay quiet about which part was actually theirs.
| What You Used To Notice | What It Looks Like Now |
|---|---|
| Arrived underprepared for discovery | A thorough brief that misses the obvious thing about the account |
| A proposal that buries the point | A well-structured proposal the seller can't summarize in a sentence |
| Vague answers about the buyer's outcome | Precise answers the seller can't source or defend |
| Hesitates to reach out | High outreach volume, no follow-up thinking |
| Wrote a weak first draft | Nothing. There is no weak first draft to see. |
Pay attention to that last row. The rough draft used to be where a manager could watch somebody think. It has mostly disappeared from the record.
Most sales organizations keep some version of a coaching log. A manager notes what they observed, works out what skill it points to, and coaches to that. If you keep one, the symptoms in it were written for work that people produced themselves, and several of those rows have gone quiet.
Here are the ones worth adding. The skill column describes the gap in plain terms rather than naming a framework, because the skill is the point and every organization names it differently.
| What You Can Still See | What It Probably Points To | The Skill To Coach |
|---|---|---|
| A thorough brief that misses the obvious thing about the account | The research was generated, so nobody really prepared | How to prepare for a conversation: who is in the room, what you already know, and what you still need to ask |
| A clean proposal they can't summarize in one sentence | Structure with no argument underneath it | How to build a story around the outcome the customer wants, rather than a list of what your product does |
| A solution that reads well and doesn't answer the concern raised in discovery | Nobody connected the concern to a capability | How to map what the customer said they needed against what you can actually deliver |
| A strong meeting, and then the deal drags | No outcome has a date or a consequence attached to it | How to qualify properly: whose problem this is, what happens if it goes unsolved, and by when |
| High outreach volume and very few replies | Volume replaced relevance | How to tailor outreach to one specific outcome instead of to a segment |
Every one of those skills has a name in whatever methodology your team already uses. At Baker Communications we teach them as Outcome Discovery Preparation, the EPIC Outcome Story, Solution Mapping, and the OUTCOMES qualification framework, and we're glad to walk anyone through how they connect. But the names matter far less than the underlying point, which is that these gaps haven't disappeared. They still point somewhere specific. You just have to look somewhere new to find them.
If you've been through any leadership program in the last thirty years you've probably met GROW. It's a coaching structure, and what makes it useful here is that the whole thing is built out of questions rather than out of reviewing somebody's output. You walk a person through four stages, in order.
Read that list again with AI in mind and you can see why it survives intact. Not one of those four stages requires you to read what the person produced. It's a conversation, and a conversation surfaces reasoning whether or not the reasoning left anything behind on paper.
Reality is the stage now carrying the most weight, because it's where the thinking becomes visible. Three questions are worth adding to it, aimed squarely at what the document no longer shows you:
Feedback is a different tool from coaching, even though the two words get used interchangeably. The version most teams are taught is SBI, which asks you to describe three things in order.
What makes SBI work is specificity. Telling someone they need to be more prepared is an opinion, and they can disagree with it all day. Telling them that on Tuesday's call they couldn't answer the question about the renewal timeline, and the CFO went quiet for the rest of the meeting, is something a person can actually do something with.
Which is exactly why AI complicates it. SBI needs a situation you were in and a behavior you actually observed. If the behavior you want to talk about is a judgment call, the finished document is no longer the place you saw it happen.
So the observation point has to move. Rather than reading the work and reacting to it, you sit with the person while they walk you through their choices, and that walk-through becomes the situation you give feedback on. It sounds like a small adjustment. In practice it means feedback on someone's thinking now takes a scheduled twenty minutes instead of a read and a reply, which is a real cost and worth planning for.
This is worth taking seriously, because the failure mode is quiet and hard to reverse. Asking a seller to account for their choices can easily land as an inspection, and once someone has decided that is what these conversations are, you will get carefully defended answers instead of honest ones for a very long time.
Three things keep it on the right side of that line.
Take your coaching notes on one seller and look at what you logged in the last two months. If the symptoms you wrote down are all about outcomes rather than about how the work got made, your visibility has narrowed, and the notes are telling you so.
Then schedule twenty minutes with one seller who is doing well. Ask them to walk you through a recent piece of AI-assisted work: what they kept, what they cut, what they checked. Say what you would have done differently.
Starting with a strong performer matters. It establishes the conversation as development rather than as something that happens when you're in trouble.
Coaching didn't get harder. Fewer skill gaps announce themselves in the work, so a manager who waits to notice something waits longer than they should.
The questions still work. They just have to be asked on purpose now, on a schedule, including with the people who look like they don't need them.
Up Next: What Identity Hours Actually Sound Like
Underneath all of this sits a harder conversation, and it has nothing to do with skill.
When part of the craft someone was proud of gets automated, the question they're carrying isn't how do I get better at this. It's whether being good at this still means anything. More than half of people using AI at work think it makes them look replaceable, which is not a training problem and won't be solved by a framework. The next post covers what that conversation sounds like when it goes well, what to say when you can't promise anything, and why most leaders skip it.
Read the Next Post →Sources
Numbered citations throughout this article draw on the source above.
