psa: this acc is just a collection of mental notes and streams of consciousness so i have receipts to say i told u so in the future. don't take it too seriously thanks :)
We've been trying to understand what can make something defensible at the AI application layer. Data network effects, specifically "System of Record" ones, are one potential characteristic. This @nickgrossman post explains further.
idea I wna call out: the notion of two scaling models involving“training” vs “thinking”. I was under the impression we’d hit a wall (since most LLMs have kinda scraped the entire internet atp), but that's on the "training" end. I’m intrigued by the idea of allocating more computing power to “thinking” which will enhance how models tackle complex problems
neobanks are succumbing to the tradfi pitfall of valuing customer conversion as a success metric-- which encourages poor kyc practices and makes it super easy for fraudsters to get onboarded
AI voice agents are on 🔥
We're moving from the innovator -> early adopter part of the curve, with new startups sprouting up weekly to serve different verticals.
What @illscience and I are seeing @a16z, and why we're excited 👇
looking at yc’s latest batch and seeing lots of wrappers on top of foundational models — lowk validates the thinking that models are commoditized and, at best, will see differentiation in memory and autonomy. definitely a higher chance of success investing in the app layer
i’m curious how these startups will fare not just in margins (from a vc pov) but also how they navigate the social sensitivities of blending business and faith
startups are pushing out “religion”-as-a-service software (online oracles), hardware (3D-printed temples), or simply repackaged social apps (religion-specific dating apps)
making a mental note of @FrichApp -- rare to see a pdt that capitalizes on gen z fomo as well as this (it’s a compliment!!) their poll feature is also a smart way to gather data for future monetization (targeted ads + partnerships w other fin products)
excited about healthcare AI startups like @AbridgeHQ — BUT i want to see it go beyond transcription to process info and propose next-steps derived from an internal medical library (for doctors to cross-check their diagnoses)