Management In The Age Of AI
Management is dead. Long live managers.
AI tools hit a true inflection point in late 2025. Building things got cheaper. AI tools got expensive. And the gap between good management and bad management got a whole lot wider.
Here’s how to think about management in 2026.
Managers Must Be Builders
Managers must be builders in 2026 for two reasons.
First, you have to learn AI tools. Without a deep understanding of how people build and execute with AI, you will be completely clueless about what to expect from your team and how to guide them.
This cannot be overstated.
A manager’s job is to do things like help people get better at their job and set expectations. Without hands-on fluency with AI tooling, you will be fundamentally unable to do these things well. It would be like trying to manage a software team in 2015 without knowing how to use the internet.
Second, building is often the most efficient path for the team. In the old world, a manager might spend five hours in meetings defending the team’s time against a drive-by request from another team. In 2026, the manager should just spend an hour and build the damn thing.
Managers must build because not building is now the bigger waste of time.
Managers Must Expect More
People have incredibly powerful tools at their disposal and companies are spending real money on them. Managers have to raise expectations of output. This will require willful effort, but that effort must be taken.
If you need a cheat sheet for raising expectations:
The trickle test. There is no longer any excuse for people not to have a steady stream of small things getting done alongside their main work. Bug fixes, small improvements, documentation — these should be flowing constantly.
Partial work is no longer partially acceptable. Whether it’s competitive intelligence, scanning the ticket backlog for signals, or writing tests — the “I ran out of time” excuse doesn’t hold when your tools can do the grunt work.
Outcomes over output. People need to add value to the business in ways that are obvious. More than ever, individuals must be end-to-end owners of business outcomes, not just contributors to tasks.
AI tooling is also putting immense pressure on underperformers. Engineers who can’t review code effectively. PMs and Designers who are bottlenecks. Leaders who can’t adapt to change. 2026 is going to put everyone below the bar underwater, and you need to be ready to step in.
Managers Must Manage Budgets
AI tools are moving to consumption-based pricing, which means managers are going to have to think about how much money to invest in each individual. This is a massive paradigm shift. It’s like if you had to decide every month how good of a laptop each person on your team gets, and sometimes people run out of laptop halfway through the month.
Start thinking through these challenges now:
Should a more senior person get more AI spend than a more junior one?
What’s the maximum you’d spend per person if you could clearly see results?
What do you do when a high performer runs out of tokens halfway through the month?
What do you do when a high performer is upset they don’t get more tokens?
None of this has established best practice yet. The managers who figure it out first will have a meaningful edge.
Goal Clarity Is Non-Negotiable
With more tooling and raw power than ever, your teams need precise goals. The fuzziness that used to get figured out later during slow build cycles will now kill you in the short run.
When your team can build fast, building the wrong thing is the primary risk.
Make sure your teams and individuals have precise goals, or you’ll spend $10k per person per month on AI tools and find out it all added up to a pile of features that didn’t move the business.
Collaboration Needs A Forcing Function
Documents are getting longer. Code is getting more verbose. Toolchains are exploding in complexity. Everyone is heads-down with their personal fleet of agents, cranking out work in parallel.
This is great for raw throughput. It is terrible for coherence.
Collaboration in 2026 requires intense, deliberate focus from managers. Week by week, you will need to pull people out of their individual sprints long enough to make sure they’re all running in the same direction. If you don’t force this, your team will ship fast and end up with a product that feels like it was built by five different companies.
Hire Like It Matters More Than Ever
A mis-hire in 2026 is catastrophic. The delta between a great engineer with AI tools and a mediocre one with the same tools is not 2x — it’s 100x. One ships compounding value. The other ships compounding slop.
Raise your hiring bar now or spend the rest of the year cleaning up after it.
Summary
Management is not dead. In fact, 2026 is the year where management becomes a key differentiator between teams that win and teams that drown in their own output.
This is the year you need to adapt faster, expect more, and start building again.
Share to Twitter , Hacker News , LinkedIn
Genuinely the first take I’ve wholeheartedly disagreed with on this blog. From what I’ve seen, the “just build” ethos has led to a trail of detritus behind each employee. There’s people who have spun up their own pseudo reporting stacks and reported incorrect figures (and come to the analytics team asking them to explain why / clean up their work), people granting agents access to confidential data (where they shouldn’t have as it’s IP and the security of models is sus), and people unknowingly stand up agents using their own credentials and unintentionally expose data to others that can’t see it. We’ve even had some senior managers set up agents with access to send out emails that have acted in unexpected (bad!) ways!
That doesn’t mean that there isn’t a path forward, or a way to fix these things, but this “build first” approach is messy, and will cause more chaos in the short-term and debt in the long-term than it’s worth.
Proceed with caution.
AI indeed raises the premium on good management, but I disagree with the underlying philosophy.
competitive advantage does not come from exceptional individuals wielding exceptional tools.
Sustainable competitive advantages rely on systems, not heroics. And in effect AI amplifies whatever system already exists.
The manager's primary responsibility is not to become the best builder, but to continuously improve the business system through experimentation.
Better goals, fewer organizational defeaters, and learning from small tests will outperform demanding more output and development from the "best builders" wielding increasingly expensive AI tools.