I think this is a fundamentally flawed take. The quote taken ab asurdem means "the CEO will manage everyone" which is obviously false.
There will always be a place for mentorship, guidance, and effective allocation of effort guided by experience. Companies will test this limit at their own peril.
@lydiahallie The issue is that the update neglects to offer the context that previously this was included in subscriptions and now it's not. Framing it solely as a credit (however large) feels a bit disingenuous at best.
I think mass layoffs sound a lot more stupid when when you frame the people as next-gen AI agents. Imagine you had access to agents like this at your company:
- They automatically teach themselves and learn from each other
- Their context windows span years or decades - basically unlimited context - covering undocumented technical and relational data that informs all key company decisions
- They run with little or no oversight; the best actually go find problems and solve them autonomously
- A subset of these next-gen agents are skilled at organizing the work of other agents and upskilling them resulting in compounding increases in performance over time
yeah no let's just nuke them
Regarding @coinbase's decision to no longer have pure managers, I 100% agree with @profgalloway:
"Every generation of executives has a fever dream of flattening the organization... but the truth is, management is the connective tissue in any organization. A lot of companies are about to discover that institutional memory was not inefficiency, it was the f*cking company."
The grand irony here is that while everyone is talking about "judgement" as the new differentiator for ICs vs AI, a lot of CEOs are displaying flawed judgement at a much larger scale by eliminating people who are good at leading other people.
Google really has an opportunity to eat Apple's lunch in becoming an AI-native operating system. They have the hardware, the model, the OS, and core apps (gmail, cal etc).
I would LOVE to see Apple completely reinvent the OS for the iPhone. It's the biggest opportunity to transform users' experience with @Apple and one of the first major leadership tests for John Ternus.
As a result of so much now being buildable, I think we are going to see massive investment in both research on users and outright simulation of users in order to speed validation. Shipping fast will always be valuable, but tooling that enables deciding fast is rapidly becoming very valuable.
Think:
1. Frontier models building AI users based on /computer-use
2. Meta building AI users based on social graph
3. AI product research agents deployed to stub roadmap based on user feedback across all channels.
The release of Symphony and /goals show how rapidly we are abstracting away from the code writing layer for engineering. Defining the end point and the system for AI to get to that endpoint is where the alpha is (for a few months at least).
My bet: being spiky at that kind of work - building and iterating the systems that make a digital team more performant over time - is going to be one of the most valuable things companies can hire for.
Every IC running agents is already a manager. The team is just digital rather than human.
Setting objectives, scoping, unblocking, review โ the job is the same. Managing yourself is the floor now, not the ceiling. A lot of folks have already observed this.
I've been building a multi-agent orchestration tool for my own work; the goal is to stop nudging four or five agents at a time and spend that time on *what* needs to get built and *why.* It's the same instinct a manager has when they invest in process or tooling instead of doing the work themselves.
I'd add that some/most UI can be removed, since it existed to enable humans to formulate requests that can now be executed in natural language.
Customization layer will by default cover smaller surface area, and will in part re-expose controls that are otherwise being handed off to agents.