Recently, ClickUp joined the (very large) club of companies laying off tons of people and trying to say that it’s “bEcAuSe Of Ai.” I had this sitting open for a while because there’s a lot of stuff I wanted to comment, and while I was well underway of a Tweet (never X) thread (EDIT: linked the thread), I realized I was basically writing a blog post in umpteen parts, so I’m just going to roll up the same commentary here to make things actually readable.
Before I start, the whole thing seemed like it was (at least partially) LLM-generated, which fits in the general theme of LLMs slopping out more means you can get more crap for the price of 1. That said, it’s not really helping ClickUp’s cause in this case.
First and foremost, the “100x organization” , which
is only possible with 10x people that have embraced and adopted new ways of working.
“10x engineers” are already a myth, claiming 100x engineering should require you to be drug-tested immediately.
AI makes the best engineers wildly more productive, and everyone else using AI slows these engineers down.
After having these tools being mandated at work, my experience is more along the lines of the more deeply you know and understand the code at hand, the bigger the gains from prompting. At that point, you’re writing prompts that have context built into them in a way not captured by the text itself – it’s in how you phrase things, the questions being asked, and the information you can give. You basically already know what’s going on, and are just looking for the specific details that fills in the last few blanks.
That’s not an LLM driving productivity or giving juniors the “skills” of more senior engineers. Getting to where you can see meaningful gains from prompting is the result of time, work, and experience. You know, the things that make you good without relying on hyper-expensive autocomplete.
If your best engineers are spending time reviewing other people’s code, then this is inherently an inefficient bottleneck. These engineers can review their agent’s code much faster than reviewing human code.
There’s probably a longer ramble on peer review in general here, but I do think it’s a good use of the best engineer’s time to review code (not all the time, but you do want it to be happening often). This is a key part of “mentoring more junior devs” that’s always a part of senior-level job descriptions.
I suppose you could just let juniors flounder around helpless and see how the whole “survival of the fittest” thing works out, but I promise you – your code is nowhere near being the “fittest.”
I say this as someone who (like many devs) doesn’t particularly like doing peer reviews, but it’s part of the job, and (when done well) offers some value.
Also, if anyone is going to review my code, I’d prefer it be someone else. The whole point is a second set of eyes looking at the code. Obviously I think my code is perfect, peer review is supposed to be a process to disabuse me of that notion by pointing out problems before the customers do.
The wrong strategy is to push every engineer to use infinite tokens. Companies doing this are celebrating 500% more pull requests. But customer outcomes don’t match the volume of code being generated.
And yet, he wants to “100x”….
Product management and design roles are merging.
There’s a general trend of AI slop along the lines of “Roles {X} and {Y} are merging” – while aspects of software development are probably overspecialized, and some more generalization is in order, this is a bit much.
My favorite variation is “Development and product management are merging” – those are completely different roles with different skillsets. The same principal applies here. It’s nice to have product teams with some basic design chops, but LLMs aren’t magically imbuing people with whole new skill trees, and pretending otherwise is not going to end well.
Everything outside of managing systems, orchestrating AI, and reviewing output becomes a bottleneck.
This emphasis on bottlenecks and removing them ignores a key fact that the post paid lip service to earlier:
The operative word in the phrase “does more s^!+” is still “s^!+.”
You must create enough disruption so that old systems are deprecated entirely. If there’s any definition for ‘AI native’ that’s what it is.
(And we all thought LinkedIn was the exclusive home of buzzwordy slop)
Nearly every company will make changes like these. The ones that do it proactively will define what comes next.
LinkedIn-ese aside, the most profitable companies in an industry do the defining for that industry. I don’t know where ClickUp falls into that scheme of thing, but doing something nearly every other tech company is doing these days is not making a case that you’re charting your own destiny.
The future is not fewer people.
(ClickUp’s certainly seems to be)
We’re already seeing entirely new roles emerge, like Agent Managers, that didn’t exist a year ago.
Literally 1 of the first things that was developed when using multiple agents was the ability to have an agent “manage” them. If anything can be done by an agent, it’s managing agents. What’s being cited here is literally the most outsourceable LLM-related work there is.
I’m not sure what the point of this original post was, but this doesn’t read like a company changing the game, how things are done, or really anything other than headcount. This isn’t defining anything.
If there’s any evidence pertinent to the question of “are any of these companies truly innovating thanks to AI?” – it’s probably the fact that this reads closer to a form letter we’ve all seen before than a vision of the future.