Review: Insights from 112 YC AI startups (W23)
Over 40% of the entire batch are AI/ML-related.
YCombinator (or YC), if you’re not familiar, is one of the oldest and most prominent startup accelerators in the world. Companies like Airbnb, Dropbox and Doordash can trace their roots back to YC.
The program “graduates” hundreds of companies per year, with both a summer and winter batch. And while YC companies are a small sliver of the startup ecosystem, they offer a glimpse at where the startup ecosystem is focused.
Right now, that's AI. Of the 266 companies in the Winter 2023 (W23) batch, over 40% are related to AI or machine learning (ML). Here are some of the key takeaways:
Lots of developer tools. YC has always filtered for technical founders, and many of them are using AI to do what they know - improve the software developer workflow.
Lots of competition. There were many companies that seemed quite similar, either in their product or their use case. Some examples: AI-generated sales emails, AI-generated code, ML model hosting/infrastructure all had multiple companies with very similar descriptions.
These companies seem very new. While most YC companies publicly launch during the program, there was a striking number who have only gone live in the last month or so.
It's unclear how defensible some of the technology is. I haven't checked under the hood, but some companies seem like thin wrappers around OpenAI's ChatGPT/GPT-4.
That said, there are still opportunities for ChatGPT competitors. Large enterprises, healthcare, and government are not going to send sensitive data to OpenAI. This leaves a gap for startups to build on-premise, compliant GPTs for these verticals.
Some of the most interesting companies are bringing AI to specialized niches. There’s AI for hedge funds, for construction projects, for drug discovery, and more. I can't wait to see what the future holds.
The full list of 112 companies is below. Be aware that the bullet points come directly from the startups' marketing materials.
Artificial Ignorance is reader-supported. If you found this interesting or insightful, consider becoming a free or paid subscriber.