As an early-stage startup aiming to build a venture-scale business, being data-driven and validating the problem space is far more important than chasing buzzwords. AI and agents are just buzzwords if you're not solving a real, painful problem. In that case, nothing is going to work. Setting the right metrics, deadlines, and timelines, while validating different approaches to solving the problem, is far more important than anything else. The challenges of scaling will follow after that.
At a stage where it seems like product is moving in the right direction. Seems like there's a strong market pull, but you never know until there's monetary traction.
I vibe-coded my own PDF compression library to replace a commercial library I paid $20K for.
After weeks of testing, I ended up rewriting most of it and honestly way better than the so-called "enterprise-grade" thing I bought.
Claude code didn't magically build it for me. I still had to read the PDF spec, trace edge cases, and understand every line. The model just gave me a baseline fast enough to iterate.
And no, this wasn't a one-shot "prompt -> shipped" fantasy.
@levelsio This is gold! What do you think is the best way to get feedback from customers if you have their name and email? I'm finding it hard to get a response from emails.
We will make the new 𝕏 algorithm, including all code used to determine what organic and advertising posts are recommended to users, open source in 7 days.
This will be repeated every 4 weeks, with comprehensive developer notes, to help you understand what changed.