โSeems like AI slopโ is now a reporting option on LinkedIn. ๐
Weโve made it so easy to sound like we have something to say that having something to say is becoming surprisingly rare.ย
We all seem to recognize slop without ever really defining it. You start losing interest before you can quite explain why.
I think weโre reacting to language that sounds like someone has done the thinking, but gives us very little evidence of it.
As a VC, when I hear a pitch, I want to understand what a founder has noticed that others missed. What a customer said that changed their mind. What they believe that someone else might reasonably disagree with.
Same here on X or on LinkedIn. I follow people for their judgment, their experiences and their slightly uncomfortable opinions that theyโre willing to put their name behind.ย Let's stop #AISlop
@ycombinator backed Pocket says it crossed $100M in arr. I had to see what the hype was about, and whether it could solve a very specific problem for me.
My meetings in TLV are in Hebrew. My investment notes are in English. My brain is usually doing both jobs.
Testing Pocket this week to see what it can take off my plate
Will report back on what works, what doesnโt, and what gets lost in translation.
American VC in Tel Aviv problems ๐บ๐ธ๐ฎ๐ฑ
We did the exact same thing with our pet insurance - except we used Instinct.
What stood out was the follow through - I was never waiting for a response from the insurer. It handled everything you mentioned, then kept chasing unanswered claims and following up until the money actually landed in our account.
Getting the claim together and sending it is one thing. Never having to chase down the insurer that doesnโt address all the claims, that is the game changer
I am sure Opus 6.x will blow our minds, but there is so much more that can be achieved prior to that with "just" better engineering instead of more capable models.
Case in point: Claude Code 2.1.318 ships a (currently hidden) Code Mode: instead of calling tools one at a time, Claude writes a script that runs them, with loops and parallel calls, and only the results come back.
(The pattern already exists in plugins. The first-party version is interesting because hooks and permission checks still run on every call inside the script)
Once turned on, we'll get cheaper, faster, longer-running agents with the existing models. Fewer round trips, less noise in context.
Binary teardown ๐ https://t.co/cmHuazcg4r
Having the capital to do more and the discipline to do less says a lot. ๐
On the geography front: SF, NYC and TLV have always remained as the strongest concentrations of founder talent in my view. Hardly a new thesis - Weโve been making the case for the latter here in Tel Aviv for a while now ๐
@_catwu You literally helped shift an entire industry!! Congrats on everything youโve built Cat. There is much to be proud of and I canโt wait to see whatโs next
First meeting, agreed. But I wouldnโt say never. If the fund is serious and wants to meet the rest of the founding team, sure, why not?
I just wouldnโt pull the whole team away from building before you know thereโs real interest. The CEO should be focused on fundraising, it doesnโt need to be a team outing ๐ Even if you need the CTO to go deeper into the tech, Iโd save that for meeting two once youโve validated the fundโs appetite.
Thereโs a right time. Just not on the first date.
I've officially converted to WhatsApp, which makes sense given the ecosystem I'm in.
But Israeli founders, hear me out: start using iMessage.
The difference between getting a response from a customer, investor, or anyone in the US within minutes versus waiting days could literally come down to the app you're messaging them on.
The problem isn't your outreach. It's the medium.