Hi @therealwolf, I'm also building AI agents, starting with one for Hive and now another for Farcaster.
To address concerns about spam and Hive's stance on AI content, I limited it to a community that accepted an AI mascot for comments.
I later further restricted interaction to followers only to keep it fun, not spammy.
Seeing how quickly HiveusMaximus was rejected, these concerns have resurfaced, but I stand with you and believe we can manage this by focusing on communities that embrace this new era of AI agents.
Hey @vaipraonde,
that's great to hear! Farcaster is much younger than Hive but gained quite a lot of traction. I've tested it myself but probably joined at a time where it was satisfied from a user-bases and the following algorithm made it quite difficult to gain traction. Which is why I was excited to build the AI agent on Hive.
For interactions, I've also focused on followers and mentions for now. It's interesting how much the output is defined be the prompt though, meaning, you can pretty much shape the ai however you want and it might be interesting to make a standard publicly visible.
Anyway, I appreciate your kind words and I'm glad I'm not alone in my endeavour to build AI agents - even though they're really just glorified API wrappers at this point, but still lots of fun to experiment with :)
yeah, always good to know we are not alone! maybe we can share some XP.
i build it using langchain, but now i want to start using eliza as well and learn how to build on it and take some advantages on those tools.
latter we can do some experimental agent-agent conversation and see what happens.