Neynar

Celebrating the experimental launch of hubs.neynar.com

is anyone traversing farcaster conversation threads and summarizing them to pass as context to their agent?

if yes, we’re building something to make this a lot easier; would love to get feedback 

reply to this thread or DC me

following up on last week, here's initial take on conversation summarizing: https://docs.neynar.com/reference/lookup-cast-conversation-summary

*pass a thread hash and prompt - get back a summary*

instead of traversing conversations, summarizing and then passing to another agent as context - do it in one step

looking for feedback, reach out if you use it

h/t @rafi who's just getting started 🪐

Engineering @neynar. Amusement, wonder, and uncertainty ✴︎ libred.org

working @ /neynar   🪐  casting @ /rish

wowow this live? i was currently fetching the whole thread myself ill try how much it changes, i guess it will be useful for the agents to use both, and get more details only if needed

CEO and Founder of $BLEU

Developing infinite memetics AI Agent @elefant


lebleuelefant.com

will test it out, what I liked from passing the whole thread convo previously to the agent's context was that it knew who said what in a structured way and could TLDR me

wondering if summarization will keep authors in context gotta try it now!

frame developer, web3 developer, long distance runner /pixelnouns /frm /degendogs /degenfers /toka /maxis

I think its fine if the reason-act agent is capable of knowing it has two skills and can use both depending on if needs more context or not to reply, but certainly some times is easier to provide more context to be safe than try to cut it back

This is my initial question too. Having the context of who said what, which are previous replies from the AI agent and which are not, etc., can be important for some use cases.