recently realized that building a new language model for crypto means enabling non ai engineers to be able to build ai agents without knowledge of fine-tuning or training. 

theres no point in trying to explain what problems / use cases what this enables. only those engineers who are building agents understand.

building in AI for crypto |
cofounder at rpslabs.io 

ex-stanfordGSB | orangeDAO fellow

100% - Mass adoption comes when it no longer feels like the bleeding edge

I'd argue there's nothing simple about RPCs or indexing, but we're working on it 😉 For better or worse though, the approach we (and others) generally take is abstracting away the complexity and nuance

Any thoughts on what would make AI more approachable / usable in web3?

exactly, 

can't expect blockchains to just "decentralize AI" while AI is still gatekept by deep domain expertise. if regular devs cant build on it, who are we decentralizing it for? 

we need to make fine tuning AI models easier so any developer can build reliable AI applications. 

as simple as RPCs or indexing