Current challenge:
- where we can get data from
- how do we use the data to power the things they are building via llm tech, either prompting or controllable programs (hard code)
social graph → interest graph (current iterations of social platforms) → taste graph!
do what spotify did with music!
all embeddings are on openai or cohere
something next: they have the data for places, llama3 is not enough to create engaging experiences than google map.
power users already love it, but pmf is small: moving to social will be the next steps: how can we recommend users to users!