vrypan |--o--|
@vrypan.eth
I'm building this. It's a LLM summarized version of my timeline on a specific day. One page per day. The UX will be much better when I convert it to a miniapp, and links will just open in the main pane. Needs work, but I'm curious: 1. Would you use it? 2. Would you pay (ex, $5/month) to have your own version? 3. What you see in the example, only shows 30 fids I follow. If I raise this number to 100 or 1000, the page will become huge. Not a technical problem, but maybe it's too overwhelming and doesn't actually provide value? https://cyan-organisational-reindeer-861.mypinata.cloud/ipfs/bafkreidhh3fbbran3e7jbiqatfxehksubvyulucp5pj6hux3kgkmxuhbdu
5 replies
13 recasts
37 reactions
shazow
@shazow.eth
Would love some "This week in Farcaster" newsletters, to catch up on any memes/drama I missed. Dunno if it's worth $5/mo to me, maybe a $25 for a year kinda thing?
1 reply
0 recast
0 reaction
vrypan |--o--|
@vrypan.eth
The main reason for charging if I make this a public service is the costs associated with punching the data and using an LLM API. That's a lot of llm tokens, if you think about it! There's room for optimization and I'm still not sure of the actual costs.
1 reply
0 recast
0 reaction
shazow
@shazow.eth
One idea I've been meaning to experiment: A service like this where the user brings their own API key (eg openrouter). Might be a good fit here too!
1 reply
0 recast
0 reaction
vrypan |--o--|
@vrypan.eth
Not so easy with LLMs: 1. Different APIs. 2. Even if you build wrappers around #1, the same prompt does not work in the same way. 3. It will actually cost more. I think that the average cost/user would be around $1/day for OpenAI. However, if it's under my control, I don't have to generate Dan's summary for every single user of the service, so the cost/user can be lower.
1 reply
0 recast
0 reaction