Source item

We are making a big bet at Silvia on specialized workflows. A big reason for this bet is because of what I learned in crypto. Years ago people believed in the Fat Protocol thesis, where majority of the value would accrue to the protocol layer. The exact opposite happened. Fast forward to today and people are repeating the same mistake. They believe in the Fat Model thesis, where all value will accrue to the model layer. I believe the opposite is going to happen again. Silvia is our bet to capitalize on specialized workflows and applied AI. We have built a ton of proprietary technology, including agentic file systems, improved memory, custom model routing, etc. While the large model companies are betting they can eat the stack and push into specific verticals with their general purpose models, we are betting that general models will struggle to outperform Silvia and other specialized AI applications. Let’s see what happens. $BRR