This is cool! Congrats on launch! How do you store user data? Do you write to device? Curious if there's a basic.tech x nxtscape collab possible here where you can store each user's info to their dedicated PDS
I've a thesis that how users use the internet will look and feel drastically different in the next few years. Rather than learning products (i.e., where is every button, what does each button do), we will be able to express intent on what we want to achieve, and apps / agents will inform us whether it's possible and when it's done.
I'm hosting a virtual hackathon with Windsurf based on this principle — one rule is DON'T build (traditional) UIs.
Date: June 21st, 8am PT
Location: virtual
Theme: One rule - don't build (traditional) UIs.
Focus
- What will the internet look like after traditional UI becomes obsolete?
- How will we provide context to agents and LLMs for personalization?
- What do generative digital experiences look and feel like — how does audio and video blend together to represent a post-UI internet?
Sponsored by:
Windsurf: The new purpose-built IDE to harness magic
Basic.tech: Memory layer for sharing user context across LLMs
Kisotechnology.com: Building intelligent systems to make principled decisions at scale
Prizes
Every participant gets one month free Windsurf credits. Cash prizes and credits for the winners to be announced.
We were frustrated by the slowness of the Deepseek app and website, so we decided to build a local-first version of it on top of our Basic database and sync tech. We chose the Groq Deepseek distill LLaMa 70B model because we felt it had the best balance of speed and accuracy.
This is an experimental open-source project that we threw together as quickly as we could, so please bare with us with any janky UI and bugs you face (happy to fix them as you point them out), but we personally have started using this instead of any of the other models and chat interfaces just because of how fast everything is
We hope this can be a pleasant contribution to your workflow
This seems pretty cool, I can totally see the use case, especially because Firebase is clunky and takes a lot of manual work to extract/study data in it
This user account was created September 25, 2018, same day as the person asking the question. For both, after 18 months, it's the first comment on hackernews, within 2 minutes.
