Anant Jain

The Manager-of-Manager+ Problem in AI Adoption

Tech

We are living through the single greatest explosion of technical leverage in history. Engineering teams are reporting 30-40% gains in coding speed. Tools like Claude Code feel like magic.

Yet, look at the quarterly results. Are products shipping 40% faster? Are customer satisfaction scores up 40%? Is revenue growing 40% more efficiently? For most enterprises, the answer is a hard no.

This is the modern productivity paradox. We have "intelligence abundance" at our fingertips, but we aren't seeing a proportionate boost in customer outcomes.

I have a feeling that the bottleneck isn't the technology. It isn’t the engineers. It’s the Manager-of-Managers+ (MoM+) layer.

In my opinion, Engineering Management as a profession went too far in considering it acceptable for MoM+ layer (Directors/VPs) to stop writing production code. This is coming to bite us hard in 2026 since one has to get their hands dirty to feel the ongoing seismic shift in software engineering[1]. As a result, far too many folks in MoM+ layer haven't reset their worldview. They are trapped in a "2023 Operating System", still managing for scarcity in a world of abundance.

The "2023" Trap: Managing for Scarcity

Until 2023, engineering hours were the most expensive, scarce resource in a company. The entire “modern” engineering management OS was built to protect that resource. In the 2023 mindset:

  • We obsess over estimation. We spend weeks grooming backlogs to ensure expensive developer time isn't "wasted" on the wrong task.
  • We write heavy specs. We demand PRDs and Technical Design Docs before a single line of code is written because rewriting code is costly.
  • We scale linearly. "To do 2x the work, I need 2x the headcount."

The MoM+ layer exists to minimize risk and predict timelines. That works when code is expensive. But hey — the code isn't expensive anymore :)

The Reality: Intelligence Abundance

Today, the cost of generating code has plummeted to near zero. In this new reality, the constraint is no longer "fingers on the keyboard". The constraints are context, taste, and decision-making agency..

The builders know this. The CEO's of successful companies operating in "founder mode" know this. The "Frozen Middle" layer of management misses this. They see AI as an efficiency tool—a way to clear Jira tickets faster. They should be seeing it as a leverage tool—a way to explore three different architectural solutions in the time it used to take to debate one.

So, as a CEO, what can you do about it?

If you are wondering why all those tokens you bought is just resulting in a lot more chatter and excitement, but ~flat growth in customer outcomes, you have to dismantle the 2023 scarcity mindset in your middle management. Here are a few ideas to get you going:

  1. Builders and sellers only: Expect everyone in your company to be managing a team of agents in addition to themselves / their team / their org.
  • The Shift: The days of "pure people managers" have been on the way out for a while, and 2025 feels like a turning point when it's potentially become extremely hard to justify your presence if you're neither a builder or a seller (preferably, you can be both in one).
  1. Restructure for "The Diamond": The traditional "Pyramid" team structure (many juniors, few seniors) is dead. AI can do the junior work. The bottleneck has moved to the architecture and review phases.
  • The Shift: You need a "Diamond" structure. Hire fewer pure juniors and more Senior/Staff engineers builders who can wield AI to own their entire vertical slice, from idea to implementation. You need fewer people writing code and more people with the taste to know what code solves customer problems.
  1. Kill the "Spec-First" Culture: Stop rewarding managers for beautiful documentation, engineering strategies, and perfect roadmaps. Start rewarding prototypes that users can touch and feel. Make it a policy: for any new feature idea, the first review isn't a document read-through; it's a demo of a working prototype.
  • The Shift: Force the organization to think in working software, not abstract plans. When code is cheap, "building to think" is cheaper than "writing to think."
  1. Rewire Your Metrics: Stop asking "How long will this take?" That is a scarcity question. Start asking "How many iterations can we test this week?" That is an abundance question.
  • The Shift: Move the goalposts from Output (features shipped) to Outcomes (iterations per successful solution). If your team ships one thing that fails, that's a failure. If they ship five variations and one wins, that's success.
  1. Watch for the “immune system” kicking in: Your middle management likely feels threatened in this new world. If their job was "managing capacity," and capacity just became infinite, they will retreat to something else ("due process?") to prove their value unless you inspire and incentivize these (usually, exceptionally driven) folks to rewire their thinking.
  • The Shift: Ruthlessly question every singe process: both proposed and existing. Think about how you can leverage the strengths of these leaders in this new world.

The next 100x company won't just have 10x better engineers. It will also have 10x better “management” that has the courage to stop managing for the past and embrace the Intelligence Abundance.

[1]: If your skip+ hasn't written a line of production code with Claude Code/Cursor in 2025, my personal advice to you is to look for a new role in 2026.