Two engineering management pieces landed in my inbox this week, and they disagree in a really useful way.
LeadDev's newsletter ran Long live the engineering manager by Liz de Moll. Every few months someone declares the EM role dead, and her case is that it isn't. AI can take over the status updates, but the coaching, alignment and translation work is growing faster than AI can automate it. Her line is a good one: org charts flatten, but dependencies don't. She saw it first-hand on parental leave, when her engineers ended up coordinating instead of building.
Then Gergely Orosz's Pragmatic Engineer ran The revised rules of engineering management with Will Larson. Will is CTO at Imprint, and his message to middle managers is much blunter: re-skill or retire. He thinks the engineering leadership profession as we know it won't exist in three years, and he stepped down from a team of 250 to one of 25 to get closer to the work.
So is the EM role safe, or precarious? Probably both, depending on which version of the job you're doing. But what struck me is what they have in common, and what neither piece quite gets to.
What's changing
Will's argument is that the old rules of engineering management were shaped by the zero interest rate era, when hiring and retention were the big differentiator. If a manager couldn't use technology to solve a problem, they had process, scheduling or headcount. His view now is that hiring is "the least interesting way to solve every problem".
A few of the shifts he describes:
Hands-on again
EMs are expected to be close to the technical detail. The old advice to never make a technical decision again is, in his words, outdated.
Bigger spans
Gergely notes that EMs are starting to manage 12 to 14 people instead of 4 to 8.
Tech lead managers
The tech lead manager role is in demand, because organisations want managers who can build as well as lead.
Painful bureaucracy
He says the tools for line management and writing software have got really good, but the tools for bureaucracy are still pretty terrible.
Swipe or scroll for more
Put simply, EMs are being asked to do more technical work, for more people, with less time. LeadDev's companion piece on the unseen work of engineering management says your best work goes unnoticed, and I'd guess that gets more true as the span grows.
The line that stuck with me
The part of Will's conversation I keep coming back to isn't about org charts. It's this: engineers are being more productive, but enjoying it less.
He mentions someone who decided to retire from tech because AI has destroyed their ability to be "in the flow". He also talks about a sense, among a lot of talented people, that the industry has betrayed them and that they're owed something for how hard they've worked. His response is that the rules have changed, and that they're right to feel that way.
You can agree or disagree with the politics of that. The practical point is that a lot of people in engineering teams right now are feeling something quite significant, and it won't show up in a velocity chart. LeadDev's Engineering Team Performance Report says trust in engineering metrics keeps falling as AI moves the bottleneck from writing code to reviewing it. That's a hint that the numbers are getting less reliable at the same time as the people are getting harder to read.
The human work is the work that's left
If AI is going to handle more of the status updates, the dashboards and the first draft of the code, the part of the EM job that stays human gets more important, not less. Someone has to notice that a good engineer has gone quiet. Someone has to find out that the team is shipping more and feeling worse.
And that is where I think the two pieces meet. Liz is right that coaching and alignment don't automate. Will is right that the old, process-heavy version of the job is under pressure. What's left is the bit that needs a person to sit with other people and ask how it's going.
The trouble is that it's harder to do that well when you've got 12 or 14 people to look after, and when the hands-on work is pulling you back towards the code.
Where PETALS conversations fit
This is the gap we built PETALS for. A PETALS snapshot is a short, regular check-in on how the team is feeling across Productivity, Enjoyment, Teamwork, Learning and Serenity. The score is only the start, because the heatmap shows you where to have the conversation.
For a stretched EM, that does a few useful things:
It surfaces what dashboards can't
"Productive but not enjoying it" shows up as a gap between Productivity and Enjoyment, long before it shows up as a resignation.
It makes a large team readable
With 12 people, you can't have a deep 1:1 with everyone every week. A snapshot gives you a team-level picture and tells you whose door to knock on first.
It gives people a safer way in
It's far easier to say "I scored Serenity a 2" than to open a conversation with "I'm struggling". The heatmap does a bit of the work of getting started.
It creates a record over time
If you re-run it, you can see whether the changes you made actually helped, rather than relying on a feeling.
Swipe or scroll for more
None of this replaces the conversation. It just makes sure the conversation happens, and happens about the right thing.
Why EMs are in a good spot to build psychological safety
Will tells a story about his time at Yahoo. In two years he had two 1:1s with his manager. In the first, the manager asked him about someone else's performance. The second was the one where he handed in his notice.
That's an extreme example, but it shows the point. People tell you the truth in places where it's safe to, and the EM is often the closest person to the team who also has the authority to do something about what they hear. That's a rare combination. A director is too far away, and a peer can't change the plan.
Psychological safety isn't built in a workshop. It's built in small, repeated moments where someone says something slightly risky and it's received well. We've written about why psychological safety makes feedback land, and about how team size changes who speaks up. The second one matters even more if spans are growing, because the bigger the team, the easier it is for the quiet voices to disappear.
A regular team health conversation gives the EM a way to make those small moments routine. Each time someone raises a low score and the team gets curious instead of defensive, the next person finds it a bit easier to speak.
A small thing to try this month
If you're an EM, or you work with one, you don't need a big programme.
1Ask one question you can't get from a dashboard
Something like "what's making your work harder to enjoy at the moment?"
2Do it with the whole team
Not just in 1:1s. Seeing the spread of answers is half the value.
3Act on one thing, visibly
It doesn't need to be big. People trust the process when something changes.
4Do it again next month
Safety comes from repetition, not one good session.
Swipe or scroll for more
The EM role may well change shape over the next few years, and Will could be right about how fast. But whoever ends up doing the job, someone needs to be looking after the humans. That bit isn't going anywhere.
Sources
Long live the engineering manager
Liz de Moll, LeadDev
The revised rules of engineering management with Will Larson
Gergely Orosz with Will Larson, The Pragmatic Engineer
Quotes and figures are from the write-ups of each piece, and the views on PETALS are our own.
This is the kind of thing our newsletter subscribers hear about first. If you'd rather get product updates and insights like this straight in your inbox instead of waiting for the blog, subscribe below.