"I Know It When I See It" Doesn't Scale
What AI taught Hilary Gridley about becoming a better manager
You can listen to the full WorkLife episode anywhere you get your podcasts or watch it on YouTube.
The more I use AI, the more convinced I become that it is exposing the places where managers have always struggled. Not because the technology is magical, but because it refuses to read your mind.
Humans are surprisingly good at compensating for vague instructions. Or at least trying to. They fill in gaps. They infer what you probably meant. They learn your preferences over time. AI is much less forgiving. If you give it fuzzy context, unclear standards, or a vague definition of success, you will just get mediocre work back.
Hilary Gridley said it on this week’s episode of WorkLife: one of the most useful things AI can do for managers is force us to answer a question we should have been answering all along.
What does good actually look like?
Hilary gave a simple example.
She found herself giving the same kind of feedback over and over. Someone on her team would draft an email to an executive announcing a change in plans. They’d send it. A few minutes later, the executive would reply with a version of, “Have you really thought this through?” Suddenly a five-minute update became a week-long derailment.
As a manager, Hilary knew how to coach people through these situations. She knew what made executive communication effective. But when she decided to build an AI tool to help her team improve those emails, she ran into an unexpected problem.
She couldn’t just tell the AI, “Make this better.” She had to explain what “better” meant.
She pulled together examples of emails she thought were strong and examples she had heavily edited. Then she asked AI to help identify the patterns. Eventually, she distilled her thinking into a handful of principles: the criteria she was unconsciously using every time she reviewed someone’s work.
The most interesting thing here isn’t that she built an AI-writing tool that her team could use; it’s that using AI forced her to make something explicit that had always been implicit.
If you’ve worked long enough, you’ve had at least one manager who said, “I need to see every email before you send it to the CEO.” It’s obnoxious, and it’s honestly bad management. One thing I think integrating AI into workplaces will make clear is that the job of a manager is not to own quality or act as a filter; it is to create leverage.
If the only way your team can produce great work is by routing everything through you, then you haven’t scaled your judgment. You’ve just made yourself the bottleneck.
The best managers I’ve worked with don’t just give feedback. They teach people how to think. They explain what they’re looking for, what tradeoffs matter, what success looks like, and where the quality bar actually is. Over time, people stop needing the manager to review every decision because they’ve internalized the manager’s judgment. This is actually one of the key Give Away Your Legos points—your job is to make yourself irrelevant so you, and the team, can scale. Part of the way you do that is by teaching people how to make decisions and produce great work without you.
Hilary’s point is that AI is an incredible partner to help you do that. It both forces you to clarify your own thinking and can act as a powerful, scaled tool to help coach your team.
Most of us carry around a lot of tacit knowledge. We know a presentation feels too long or when an email buries the lead or when something isn't quite right. But often, we don’t know how to explain why, and no one has ever forced us to.
We may have gotten to where we are because of our intuition, but the challenge—and the leverage—is turning that intuition into something you can teach.
Ironically, AI may end up making us better managers because it forces us to do something we should have been doing all along: making our judgment visible.
Your team has always needed that clarity; AI just refuses to pretend it already has it.
Want a sneak peek at my conversation with Hilary? Check out the clip below:
What else?
If this resonates with you, here are things you can do next:
Subscribe to Lessons to get new posts. I post at least 2X per month for all, and paid subscribers also get access to video courses and the full library of content.
Listen to the WorkLife podcast, where I talk to brilliant leaders and builders to untangle the messy, human side of work.
Follow me on LinkedIn, where I share more tools, thoughts, and a little of the day-to-day chaos.
And if you’re a leader looking for a community of peers to lean on and learn from, come find out more about Glue Club over here.