I've been making small very specialized toolkits to help my workflow. All of them will be made using cross-platform native desktop technology. Feel free to add any idea suggestions.
The first @ycombinator startup to visit our office! Great meeting with @gerynugrh, founder of Halo AI https://t.co/FEQWw9sIHQ
YC Summer 2022: https://t.co/7SVXQKdmzf
@dalangcloud This is correct while you're finding PMF, but once you find one, moving to AWS/GCP is a no brainer.
Cost bisa di reduce signifikan dengan CUDs, dan most AI agent nowadays bisa operate AWS/GCP cli (or terraform) hence maintenance cost almost 0
React is a bit broken, to optimize a React apps, you'll need to restructure your component hierarchy to isolate re-render.
Every optimization effort becomes a rewrite to the entire component structure.
Why is there no cloud provider that allow us to host an on demand long running agent?
Vercel limit execution time to 15 min, cf kills your runtime on every deployment.
Using sandbox does help but the dx suck.
I feel like this is a big gap waiting to be solved
@ilhamfp31 AFAIK Railway, Render, https://t.co/vgaU6nPpwU have the same problem where after you deploy your change, the previous server will be given a time of 1m before being killed, it's not doable in production workflow
@ariaghora Tauri pakai default rendering engine dari tiap OS. Hence, compatibility & behavior tidak bisa di guarantee 100% sama antar tiap OS.
Untuk app yang simpel, this is ok, but for Codex/Cowork mungkin akan ketemu quirk2 dari Webview vs Chromium
we built autonomous AI agents that hunt illegal online gambling networks in Indonesia. the agent discovers sites on its own, actually registers (!!), and extracts the financial accounts behind the operations.
take a look at our codex-powered agent navigating a live gambling site, fully autonomous, no human in the loop:
@girikuncoro @tiktokaustralia Amazing mas, what a journey. I still remember how passionate you are about CNCF when you were my hackathon's mentor.
Best of luck!
@linear Your current slack agent keeps duplicating a lot of issues.
I think your current agent see the whole message inside a slack channel when running? Instead of scoping it to a slack thread
How is youtube music recommendation so good... You start with one nostalgic song, and before you know it, it’s serving you hundreds of tracks you didn’t even realize you remembered.