Her main argument is that companies no longer change as one group. Even with the same tools and leaders, teams move in very different ways. Some rush ahead, some move carefully, some resist, and some can push too hard. But AI does not cause this, it just makes it easier to see.
In your article, you explain why this happens. Teams feel different kinds of pressure. Some are close to customers or money, while others worry about rules or risk. Some teams feel safe to try new things. Others do not. Past experiences matter too. All of this creates different “climates” inside the same company, like sunny, cloudy, or stormy weather.
Let’s jump into the discussion
Your idea is that, instead of asking, “Is our company adopting AI?” the better question is, “Which part of the company are we talking about?” Leadership should no longer be about forcing everyone to move the same way. It is about noticing where teams are, supporting them differently, and helping everyone move toward the same direction.
Here’s the question I want to open up: How do you lead when teams have the same goal, but not the same pace?
Let’s explore organizational change with Stephanie Muxfeld a VP of Product Management who has seen these challenges firsthand. I’m excited to dig into your four archetypes!
Great to chat with you this morning Stephanie. Let’s jump in!
AI has taken over our feeds and put pressure on teams to figure out how to incorporate these new technologies. To begin, let’s start by framing our discussion on how you look at AI in your role today. What shapes that view?
Hi, Bryan - thanks for having me! When I think about AI in my own role, I acknowledge that it is being shaped by things I haven’t even consciously noticed. I’d argue it’s the same situation for most people. A couple of the things I’ve been reflecting on recently:
What pressure are my customers, and my company, under right now?
What does AI threaten or enable in my work and in my customers’ workflows?
What’s my relationship with uncertainty, my company’s, and my customers’?
My view isn’t fixed, and my customer’s isn’t
either. It’s shaped by context, incentives, and what your seeing around you.
This is the core tension I am bringing to light in this article. Even when everyone agrees they want to get to the same place, some teams are springing and others are not.
There are 3 things I try to focus on in these situations:
Make the destination crystal clear. Not the timeline, not the how, just the end point.
Create visible progress markers. Teams and people like to see visible progression, and be able to tell if they are heading in the right direction.
Build bridges between the fast and the slow. Pair them up. Let the fast teams demo. Let the slow teams ask the really hard questions.
The hardest part is letting go of the idea that all teams should be sprinting at the same speed. Some teams have a genuine reason for moving slower (compliance, regulatory) and we should honor that as much as we can.
My job isn’t to be the pace setter, it’s to be the navigator. Of course pace is important, but only if we know where we’re trying to end up.
This is a great way to look at these challenges because it recognizes that things are still changing. Absolute certainty is fool’s gold.
That said, in our work, I don’t see futures that do not include the new patterns AI is creating and forcing on teams. So I believe this creates a pressure cooker.
This is a rather bold statement, as most teams have specific revenue goals they need to hit. Most of these AI investments are costly and disrupt margins. How should leaders think about this specific problem when working with executives who have mandated goals?
This is the real question! Thank you for pushing on this.
You’re right, and I did make it sound too clean. When your executive team has a revenue target and margins are getting squeezed by expensive AI tools and the productivity dip that comes with adoption, “clear destination” can’t just be “we’re AI-enabled someday”.
Here’s how I think about it when financial reality hits, or when I’m talking with executives and boards:
The destination has to include the business outcome, not just the transformation. “We’re moving towards AI-enabled workflows” is not a destination boards, execs, and investors care much about. “We’re increasing deal velocity by 20% while maintaining gross margin about 75%.” That’s a destination. AI might be part of how you get there, or it might not be.
You have to make the trade-offs explicit. I know this audience knows this more than most. But to give a specific example in this context, “We can hit this quarter’s revenue target if we stay on current workflows. Or we can invest in AI adoption, which will cost us $x and slow us down for 4-6 months, but position us to scale without adding headcount in 2027.” Make them choose. Don’t try to do both and quietly fail at one.
Pilot with constraints that protect the business. Test AI with teams or workflows that won’t blow up the quarter if they stumble. Let your high-performing, stable teams keep doing what works while you experiment elsewhere - instinct is to often experiment with your best teams, but this is not always the right approach. Protect your revenue engine while you learn.
PE-backed reality check. I’m at a PE-backed company, so I get it - efficient growth matters more than growth at any cost. Sometimes the answer is: we don’t adopt AI thi syear because the ROI isn’t there yet and we need to hit our EBITDA targerts. That’s a rational decision - not every company needs to be early.
Pretending that you can have transformation AND perfect execution AND margin expansion all it once is a mistake. You can’t. Leaders who succedd name the trade-off and make an intentional choice.
Agreed. And I believe it’s why some companies are valued much higher than others. Those who know how to manage these risks will outperform the market.
I love this. One of the big challenges teams face is creating a faster feedback loop. Pilots need to have an intense period of learning and adjusting, more than they ever have before. (Posting Andreas Johansson’s article this morning on the Helio channel around this topic)
In my experience, this is where teams start to fall apart: new skills force people to learn to collaborate differently. This is the start of the pressure… as everyone’s expectations are high.
How should teams think about learning and training in this new world when morale takes a dip?
We should have a separate dedicated session on feedback loops. So important and SO HARD to get right, especially give all the priorities and constraints we’re facing in the current environment. I’ll read Andreas’ article because this is an area I’m constantly trying to level up and there is always more to learn.
When morale is low, the last thing people want is to be told “and now you need to learn this whole new thing on top of everything else you’re already drowning in.” I have had several people ask me why they’d want to learn something that they are confident will process them out of a job (AI).
So first - don’t call it training. Training feels like school and homework. One more thing on their never ending list.
Try messaging it as reducing current friction. Frame it as “You’re spending 3 hours a week on status reports. What if we could get that down to 30 minutes?” That’s not training, it’s relief. Same for lo-fi prototypes - what if we could reasonable do a lo-fi prototype for nearly every feature? And we could do it in 15 minutes? Wouldn’t that reduce friction with the engineers who have to build it?
A couple other thoughts:
Make it opt-in, not mandatory. Especially when morale is low. In these cases, autonomy matters more than ever.
Pair fast learners and early adopters with burned-out humans. Don’t send people to a training portal. Let someone who’s already figured it out sit with them for 20 minutes and show them. This also helps build connection which can help in low-morale environments.
Celebrate time saved, not skills aquired. Nobody cares about a certificate when their exhausted. They care about getting to their family on time. Make the results visible: time saved, outputs maximized, prep time cut.
Acknowledge when it’s not the right time. If your team is truly underwater, it might not be the best time to add new tools. There is rarely a perfect time, but there often a better time. Protect your people.
Thank you so much for sharing this article. It is really good and reveals a harsh reality within companies today. I love the concept of microclimates that behave differently, each with its own pace. My question for you is: as a manager, how do you convey this diversity (knowledge, experience, technology) toward the same goal without losing product coherence?
In a past experience, one of the challenges I faced when adopting AI was that not all teams were open to exploring, and work dynamics started to shift into something new and great at some point. Product consistency must be maintained regardless of the internal rhythms of the teams.
This is one of the questions that keeps me up at night. Because you’re right - you can’t just let teams drift apart and hope the product stays coherent.
Here’s what I’ve learned (usually the hard way):
Separate process diversity from output standards. Teams can use totally different methods to get to the same quality bar. One team uses AI to generate wireframes, another sketches by hand. Fine. But both teams need to hit the same accessibility standards, the same interaction patterns, the same information architecture principles. The output standards are non-negotiable. The process is flexible.
Make coherence visible and measurable. Define what product coherence actually means for you. Is it design system compliance? Is it consistent user flows? Is it brand alignment? Whatever it is, create lightweight ways to measure it. Then you can see if diversity in process is actually breaking coherence or if it just feels uncomfortable.
Create shared review rituals. This is where the magic happens. If teams are working differently but coming together for critique or review, they see each other’s work. They calibrate. They ask questions. Coherence emerges from conversation, not from mandating identical processes. In my own experience, designers have always been my best role models for shared reviews!!!
Standardize the handoffs, not the work. Teams can explore however they want but when they hand off to engineering or to another team, the deliverable needs to be consistent. Clear specs, consistent documentation, predictable quality. That’s where you enforce standards.
When coherence breaks, address it directly. If a team is moving so fast with AI that their work no longer fits the product, that’s data. Pull them back. Show them where it’s breaking. Don’t let it drift. But also don’t assume it’s breaking just because it feels different. Same if a team is going too slow to keep up with the teams around them; that’s data too, and a signal that something needs to shift.
The hardest part? Trusting that coherence can emerge from diversity. But in my experience, the products that feel most coherent aren’t the ones where everyone does everything the same way. They’re the ones where everyone understands what matters and calibrates constantly. We will always govern the end result, but we can be more flexible in how teams get there (most of the time).
Yes, Bryan. This is where it gets really messy. I’m not going to pretend I have the exact answer or even that there is a right answer .
When you make it opt-in and some people sprint ahead, you create a new problem. The fast people are frustrated waiting. The slower people feel the gap widening and lose morale. And you’re stuck trying to hold it together.
Here’s what I’ve seen work (and what hasn’t):
Let the gap exist for a defined period. Give it 4-6 weeks. We don’t always need to “fix it” immediately. Fast people experiment, slow people watch. But set a checkpoint: “In 6 weeks we’re deciding - does this become standard practice or do we shelve it?” That gives fast people room to move, slow people time to observe, and everyone knows the gap is temporary.
Create AI-optional and AI-native tracks for different types of work. Some features get the fast AI-assisted approach. Others get the traditional craft approach. You’re not splitting teams permanently, you’re splitting work streams. Fast people take one track, methodical people take another. Then you compare outcomes and quality. I’m doing this with some of my product managers right now; some are very willing and excited to use AI to assist in their feature briefs, PRDs, and specs. Others not so much. That’s fine, but no one gets more time in their day and everyone is responsible for the same quality bars and timelines. I will say this is one area where I already know the answer - AI assistance will become mandatory - but I’m letting the process play out for a quarter so I have the data and signal I need to make it defensible.
Yes, sometimes you shift teams. If someone is genuinely moving 10x faster and the rest of the team isn’t ready, that person might need to move to a different squad or initiative where speed is the priority. It’s not punishment. It’s putting people where their energy can be useful instead of creating frustration on both sides. I recently did this with someone on my team - moved him to a team working only on AI features because it matched his energy, curiousity, and eagerness to have significant impact. While others on the team where disappointed they didn’t get that opportunity, this person was the one who demonstrated that they’d be a great fit for that work and pace.
Have the honest conversation. Sometimes you need to say to the team: “Here’s what I’m seeing. Some of you are way ahead. Some of you aren’t ready. That’s creating tension. What do we do?” Let them problem-solve it with you. They might surprise you. Not always, but it does happen sometimes.
Accept that some talented people will leave. This is the part nobody wants to say, and it’s been one of the hardest lessons of my career. If someone is truly 10x faster and your organization isn’t ready to move at that speed, they might go somewhere that is. That’s not failure. That’s climate mismatch.
Beautiful…and it’s why product management in some form will continue for the foreseeable future. These are human problems, not efficiency issues.
Absolutely. I think this is where leaders are really struggling. Throw in remote workers, and now you’ve got even more challenges of holding a standard.
All that said, I’m really optimistic about the technologies and how they can work magic.
I’m optimistic, too, Bryan. There is a lot of hype about AI taking over product management (and design to some extent) and I think it will have impacts for sure. But soon enough leaders will realize that you still need someone with skills and experience to guide the AI. At least that’s my hope!
You are right. This is a solid reason why designing a system, rather than focusing on isolated actions, is key for leaders: in this way, it is easier to spot subtle nuances, make better calls, and have “control” over challenges. Technology will continue to evolve, and so will the pace of teams, but having a general overview can make a difference.
All of this, the micro-climates, the pace differences, the morale issues, the adoption tolerance, it’s not an AI problem. AI is just the pressure-test that’s revealing what was alrelady true about how your company operates.
The questions in this thread are the right ones:
What’s our relationship with uncertainty?
Do we value speed or stability more?
Do we optimize for the fastest people or the collective?
Do we believe in experimentation or lower-risk development?
These questions have always mattered. AI just made them impossible to ignore or gloss over.
I would argue this is the absolute BEST moment to revisit operating values. Not in an offsite values-workshop sort of way, but in a “what do we believe RIGHT NOW???” sort of way.
Because here’s what happens: companies say they value innovation, but they punish people who try things that don’t work. They say they value autonomy, but they mandate uniform adoption. They say they value people, but they grind them down with conflicting priorities. They say they are a “product first company” but they don’t invest in product, design, or the things that enable amazing product and design.
Ai adoption is a forcing mechanism - you have to choose and really can’t get away with not.
It’s time for companies to get honest about their values and then design systems that match. If you value low-risk, carely, quality-driven work, then build your AI adoption around that. If you value speed and market leadership, own the trade-offs. Both are perfectly legimiate business strategies but you really can’t be both.
I think many of us are feeling the pain right now of the gap between what companies say they value and what their systems are actually rewarding.
Wow. This was well-written. Love everything you’re talking about, @stephanie_muxfeld!
This is where I see the old saying,
“A person is smart. People are dumb.”
being even more relevant in today’s age. Companies keep losing trust, taking shortcuts, to obtain the ultimate goal (money). It’s really hard to see, looking from within the organization, vs seeing it from the outside-in. AI will “complete the transaction” if companies keep moving in the same direction.
This also feels like there’s an implicit detachment from leadership knowing what they’re actually trying to do, or simply becoming lost in the sauce under stress.
I’m curious if you have any thoughts on how organizations can refrain from doing back-handed things, especially in the age of AI.
Hiiii @stephanie_muxfeld - I am jumping in here, coming from an Ops perspective and I have sooooo many thoughts. First of all, this Q&A is SO VALUABLE and your knowledge and perspective here is sensational, so thank you. These framing questions really stood out to me.
This is also truly SINGING! My question is.. What signals tell you a company is serious about its values versus simply narrating them?
