Saam Motamedi, Partner at Greylock
Navigating the AI Revolution: Insights for Startups
On June 13 we had the honor of hosting the brilliant Saam Motamedi, Partner at Greylock, for an insightful SV Icons Generative AI dinner. Saam invested in multiple AI companies, such as Adept, Snorkel AI, Cresta, Modular and many more. Here are some takeaways from our engaging conversation:
1. Many AI application startups are only adding natural language interfaces to existing solutions, rather than reimagining workflows and tools from the ground up. If you want to win in the space, you need to change the experience of doing things old way as incumbents are also quickly integrating AI in their products/solutions.
2. To target incumbents, startups should focus on areas where they are slow or not focused, gradually chipping away at their market share by offering differentiated and better experiences. It is incredibly hard to fight tech incumbents in their core competency.
3. Data is important and edge cases really matter, but some AI applications actually don't need a huge amount of data to be effective. Pretrained models and synthetic data can help reduce data requirements.
4. Copilots and AI assistants that fully understand users, contexts, and their personalized needs will offer the biggest potential value in the AI wave. Every single profession will have a co-pilot in the next 5-10 years. In consumer space Apple has a big advantage in this space since they have the distribution and consumers trust them with privacy issues.
5. As the landscape of large language models evolves, it's challenging to predict the future success of related technologies and businesses. Areas of interest include vector databases, security concerns, observability, quality analytics, and new tools like prompt engineering. However, it is too early to determine which opportunities will have a lasting impact and result in successful ventures.
6. There will be a significant shift in the labor market towards software-based solutions, with more companies using software to replace labor. This shift is expected to be larger and faster than in previous generations, and the price points for these software solutions may be higher.
7. The current AI startup valuations may not be sustainable, leading to potential pain and corrections as companies struggle to grow.
9. Open source models can get startups most of the way to initial product-market fit for many use cases, though dedicated models may be needed for products requiring high performance.
We'd like to extend our sincere gratitude to Saam Motamedi for sharing his valuable insights and expertise and to all the participants who contributed to this thought-provoking discussion! Let's keep the conversation going and explore the exciting opportunities that lie ahead in the AI space! #generativeai
Organizers: Aizada Marat and Chyngyz Dzhumanazarov