Content
@
https://warpcast.com/~/channel/aichannel
0 reply
0 recast
0 reaction
Mo
@meb
[Thread 🧵] - Summary of yesterday's presentation by @jrf Product vision @atlas faciliates high quality connections and meaningful social interactions Focus is on front-end agents, ie. agents a human can interact with. Technical underpinnings - Each message to Atlas relies requires a certain kind of context (user context, thread context) - Existing frameworks like Eliza or Langgraph are heavy. Having full control over what you are sending to the LLM is super important. - Atlas can interact with different modes ie. personalities. Understanding which "mode" is most appropriate for a user is a key part - Embeds can allow the invocation of Atlas in specific modes. This means an infinity of frames powered by Atlas could be created. This is a whole new design space. - Filtering measures allow Atlas to avoid spam and engage with users that have a minimum level of reputability. In future this might be more fine-grained, with users instead
4 replies
3 recasts
7 reactions
Mo
@meb
Business context - B2C is a great place for agents. Since they can now be active participants in conversations. - Agents can act as conversational mediators, prompting humans to engage and take actions that are fun to them - Farcaster is a great environment to build. Open social graph, high friendliness, very easy programmable interactions - Gaming and entertainment will be huge areas of growth - On the community / b2b front, agents can help with re-engagement of a community
1 reply
0 recast
2 reactions
Mo
@meb
Building an agentic framework - Ultimate outcome would be a declarative single config file to declare your agents and give them access to tools - The framework should handle the essentials of agentic building while freeing the user to build at the application layer - Potential direction is something that makes it easy both to spawn agents, and serve new frames
0 reply
0 recast
1 reaction
