What Top VCs are Excited About in AI and What They're Tired Of
Artificial Intelligence (AI) is still a hot topic, especially when compared to other tech areas that are currently having a tough time. Despite job cuts and startups closing, AI and machine learning startups are generating a lot of interest and money this year. But, not all AI startups are equally appealing to investors. We talked to nine top Venture Capitalists (VCs) to find out what kinds of AI startups excite them and which ones they’re tired of.
The “Picks and Shovels” of AI are More Popular Than Another LLM
Large Language Models (LLMs), like OpenAI's ChatGPT, have grabbed a lot of headlines. These are AI systems that handle language tasks using huge amounts of data to understand language patterns. But, investors feel that the foundation model layer for LLMs is now too crowded and dominated by big companies with lots of money.
Navin Chaddha from Mayfield Fund says, "This area is tough for new startups and is now overfunded." Instead, they are more excited about startups building AI’s infrastructure layer—the basic tech that makes AI more powerful and easier to use.
For example, vector databases analyze and manage large amounts of unstructured data, like text documents and images, which don't fit neatly into a traditional database.
Advanced AI on Unstructured Data
As AI evolves, unstructured data is becoming more important. Unlike structured data, which is neat and easy to analyze, unstructured data includes text documents, images, audio files, social media posts, and videos. AI algorithms in natural language processing and computer vision thrive on this kind of data, revealing patterns and insights that structured data can't.
Chaddha from Mayfield says this type of data is crucial to AI's "cognitive plumbing," the tech needed for AI applications. Chip Hazard from Flybridge Capital Partners adds that although startups working on unstructured data might not be as flashy as their LLM counterparts, they are vital for AI’s progress.
Specific Use Cases Over Crowded Markets
Some investors are moving away from general solutions in favor of vertical-specific AI—AI designed for a particular industry. Rak Garg from Bain Capital Ventures says, "I'm not interested in AI that just sits on top of tools like Salesforce." Instead, he prefers ultra-vertical AI startups that do one particular job really well.
Examples include:
- EvenUp: Uses AI to draft personal-injury demands for lawyers.
- Norm.ai: Helps banks with regulatory compliance.
However, some markets are already overcrowded. Kahini Shah from Obvious Ventures mentions that areas like notetaking, productivity tools, marketing, and copywriting are so filled with existing products it’s hard for a new company to stand out.
AI to Revolutionize Software Development
Lauri Moore from Foundation Capital believes AI will transform software development. AI can automate repetitive coding tasks, quickly spot and fix bugs, and generate code based on natural language descriptions.
A good example is DLTHub, a startup creating a library for Python, a popular programming language, specifically designed for AI workflows. Chaddha from Mayfield Fund says AI will create “AI Teammates” to assist human developers, making them more productive and creative.
Personalized AI Solutions
Finally, VCs are looking for AI startups that offer personalized solutions and actually solve real problems, rather than just hyping up AI. Foundation Capital's Moore says, “There's an over-emphasis on transformer models.” Instead, founders should start with the problem they want to solve and then choose the right technology.
She is skeptical about startups just adding an AI chatbot layer to everything. Moore also notes, “I've heard many pitches that aim to be the next OpenAI for enterprises. But, OpenAI itself is already focusing on enterprises.”
Key Takeaways
- AI "picks and shovels" (infrastructure tech) are more appealing than new LLM startups.
- Unstructured data is crucial for advancing AI capabilities.
- Investors prefer ultra-vertical AI startups over crowded markets.
- AI will revolutionize software development by automating tasks and enhancing productivity.
- VCs want personalized AI solutions that address real problems, not just generic tools.
In Summary: Investors are looking for specific, well-thought-out AI solutions that make a real difference, and they're tired of generic AI tools that don’t offer unique value.
