Square just laid off 4000 employees.
It does feel like many technology companies over-hired during the pandemic and are now blaming AI for lay offs — but at the same time, you cannot deny the productivity gains from AI and the inevitable redundancies from it.
But what do we do? Yes, I am sure there will be companies with exceptional revenue:headcount ratios due to AI, but this will be an exception because running a meaningful business is just too complicated for AI.
Engineering
I think that engineering will undergo an aggressive change in the next few years — engineers won't write code, they'll write tests and/or technical documents that fleets of agentic coding agents will work on.
An engineering manager will manage a fleet of 100 agentic coders and a handful of engineers, rather than a dozen (human) engineers.
I think it'll be really difficult for graduate engineers for obvious reasons.
Product
I am a little less concerned about the PM function because there is already a ratio of leverage between PMs and engineers — 1 PM can collaborate with dozens of engineers grouped under a few engineering managers.
It feels very unlikely that the PM function will be migrated to a complete agentic experience. I do think PMs will manage agents that are doing customer research, product analytics, project management, and such — but I am very confident that PMs will remain the "editor."
For PMs, it feels like these agents are actually replacing software. I won't use Amplitude anymore, I'll have an agentic coder that does my product analytics.
I think that using agents will become a critical skill for PMs — in many ways, I think the current PM role will be very disrupted by this shift, because leveraging agents to maximum output is a somewhat technical task.
PMs should be capable of creating skills that are specific to their function.
I wonder if this will lead to a resetting of the PM function, towards what it was a decade ago, where there was an expectation that PMs are modestly technical, rather than the current Product Management Industrial Complex.
I believe that this is a key shift because many PMs are not technical and are ironically doing tasks that agents can do really effectively — project management, program management, stakeholder updates, and so on.
Suddenly PMs are able to focus on strategy because they have agents doing those aforementioned tasks. Moreover, PMs will return to a highly leveraged role where they're capable of providing strategy for an even larger cluster of engineering managers.
I think it's very likely that a Group Product Manager who manages 2-4 PMs today can suddenly do that work themselves with agents.
But due to the radical shift in day to day work in a pre/post agentic PM world, I think the GPMs of today are not necessarily the same GPMs that will be effective in the future.
The skills are fundamentally different, and ironically a GPM from today may have significant atrophy of the actual strategy skills since they're focussed on "product management management" (man management of PMs.)
My advice
This is my personal advice that I intend on following:
PMs must become more technical
PMs must become exceptional at understanding the AI stack — from the model (processor) to the agent (operating system) to the skills (applications) — put another way, if you're not creating skills, I doubt you're a future PM
PMs of today need to move into roles where they can maximize their AI leverage. I think for the next 1-4 years, this means becoming a PM that encourages AI adoption either internally or for their product surface.
Put another way, your next PM role needs to be one where you can tell a clear story of how 5-10 other PMs can be fired because you're there. Woof!