I read this at 7am and immediately called Melinda. (yes we are up and have small 5 children between our families)
Anthropic just dropped the most important report of the year for founders and investors.
If you’re building a customer support SaaS right now, I’m sorry but you’re toast.
Labor market impacts of AI isn’t just another research paper. It’s a data-driven map of which jobs AI is eating right now and which ones it can’t touch. If you’re building a startup or investing in one, this changes how you think about everything.
Helping you boil down the report. You are welcome.
AI can do way more than it’s actually doing
The gap between what AI can automate and what it is automating is massive.
Take Computer & Math jobs. 94% of tasks are theoretically automatable. But only 33% are actually being done by AI today.
That gap? That’s your build and our investment window. The companies being built right now to close that gap are where the some juicy returns will be.
But here’s the thing, not all gaps are created equal.
The jobs AI is already eating
Anthropic measured actual Claude usage across every occupation category. The most exposed jobs right now:
• Computer Programmers - 75% of tasks already covered by AI
• Customer Service Representatives - #2 most exposed
• Data Entry Keyers - 67% coverage
• Financial Analysts - increasingly automated
If you’re building a startup that serves these workers, you need to ask yourself a hard question: will your customers still exist in three years?
The jobs AI can’t touch
Here’s what surprised me most. 30% of all workers have zero AI exposure. Zero.
Cooks. Mechanics. Electricians. Bartenders. Plummers.
These aren’t low-value jobs. They’re jobs that require physical presence, human judgment in real-time, and hands that can actually do things in the real world.
This is where it gets interesting for founders.
The O-Ring Principle
The report references something called the O-Ring model. Think of a job as a chain of tasks. If AI can do 9 out of 10 tasks but the 10th requires a human - the whole job is safe.
That one irreplaceable task is the O-Ring. Break it, and the whole thing fails.
An electrician needs to physically wire your house. AI can handle the scheduling, quoting, invoicing, compliance paperwork, and customer communication. But someone still needs to show up and do the work.
The startup opportunity isn’t replacing the electrician. It’s automating everything around them so they can do twice the work.
We put our money where our mouth is. Andre from Traderun is doing exactly this - fixing everything around the electrician. Scheduling. Quoting. Invoicing. Compliance. Customer comms. All automated. The electrician just does what they’re best at. Wiring your house.
10x better.
That’s the bet
What this means if you’re building a startup
The safest startups right now? They serve workers whose core task is physical, relational, or legally protected. Build for the hands, not the desk. AI makes these people more productive. It doesn’t fire them.
And Anthropics own data backs this up. They weight augmentative AI use at half the displacement risk of full automation. This means: tools that help humans work better are stickier than tools that replace them. Your customers stay because they need YOU!
One thing that really got me. There is no mass unemployment, yet.
But there’s a 14% drop in hiring for 22-25 year olds in exposed occupations. Companies aren’t firing people. They’re just not replacing them! If your startup’s customers are quietly shrinking their teams, your TAM is contracting in real-time. No matter what your market size slide says
The exposed workers are not who you think
This is white-collar disruption, not blue-collar.
The most AI-exposed workers:
• Earn 47% more than unexposed workers
• Are 4x more likely to have graduate degrees
• Are more likely to be female and white
The people who spent years getting educated for “safe” careers are the ones most at risk. The tradesperson who never went to university? Their job is more secure than ever.
Think about what that means for your children’s career choices.
How we think about this at Karaoke Club
If the AI labs are already doing it? Don’t compete. If nobody’s built it yet? Go go go!
Build the vertical product before AI labs ship it as a feature.
Physical work, regulated industries, deep domain workflows. By now you know we love to invest here.
I checked our entire portfolio of 20 companies against Anthropic’s framework. And we are walking the talk. Thank God!
The biggest startup opportunities of the next decade are not in building more AI.
They are in bringing AI to the 30% of workers who haven’t been touched by it yet. The ones who do real work in the real world.
If you’re building where AI meets the real world. Talk to us, we want to know more.
No posts

Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.