We started Zeed at the end of 2021, the tailend of an era that now look like you could've got funded with any idea you drew from a slot machine. Generally speaking, what it meant to be a 'consumer startup' was pretty well understood. I think that has changed now.
YC
There are super spreaders in the startup world, the people (and organisations) whos ideas have a contagious explosion to spectators, commentators and VCs looking for their next thing to talk about. YC is one of them -- and they're saying that consumer's back, thanks to AI.
For context, 2023-2024 was really an exodus of the consumer startups funded in 2020-2021. There is widespread on consesus on what doesn't work, with lots of pieces of hard evidence and data to back it up. Just because there's a (not so new now) platform and technology on the block doesn't magically eradicate most of those problems with consumer startups. I am bullish consumer, and think it is the hardest category to be operating in by a country mile, but I think it needs a clean-up on what we broadly define consumer startups as. We need to forget about the meriocratic explosion of consumer apps pre 2016 (Airbnb, Snapchat, Uber and Coinbase)
The Consumer Rug Pull
There a very few consumer startups that have, or are currently tracking towards mass scale in the past few years. And any reminant of a playbook is in tatters. It quite simply comes down to this; it is too difficult and expensive to acquire users that repeatedly use your product, are willing to pay for it and tell other people about it. Even for the ones that go viral, often aren't sticking around for too long. It feels like any form of success is just a get rich quick scheme, akin to a crypto rug pull category of products. Yes, maybe I'll pay out of fomo and to get some instant gratification but subscribe for a year? hell no! A few examples:
- EPIK (college yearbook photo app)
- Locket (homescreen photo sharing)
- CheaterBuster (see if your partner is on Tinder)
- Gas (see who likes you)
But, AI???
As a small side endevour, I wanted to test out Tavily, an internet search API made for LLMs. I took a16z's hot 100 AI companies, scraped what they did online and classified them. The clear winners are in three categories:
- AI Chat
- Roleplay
- Photo Generation
Roleplay is here to stay, the market will grow and the products will get better, and most importantly people will pay. That's great. The rest of it? Won't stick, or more precicely won't be better enough in comparison to ChatGPT to compete or justify paying for.