2 Comments
User's avatar
Tanuja's avatar

This is super cool to read, thanks for sharing! I'm curious, you mentioned you're not a native Python coder but do you have any other coding background or none at all? Just wondering how doable something like this would be for someone without a technical background. Thanks in advance!

Expand full comment
André J's avatar

Yes. I have a background in tech. But If I can do it 24h. Then it will take someone without a background in tech. Maybe 3 days instead. Its really about making it in steps. Solving one piece of the puzzle at the time separately. Like lego. Its easy to build small things .But once it gets bigger its harder to make progress. So its really important to have very limiting constraints on the goal. As the AI is great at reasoning about small chunks but like humans, gets very confused and unsure once things become more lengthy. So the trick is to stay humble on features. Its easy to get carried away, but then you lose the ship sort of. Once you make it work. Then you can get more ambitious and try to add and improve it. But thats another module and another task all together, and you will have to start small again. Shoehorning in increments of progress into the mother ship. To build out this project as a profitable business it would take maybe 2 more weeks of 24/7 work. And then build the business side of it and marketing so maybe 1.5months before it can earn a profit. Because everything gets more time-consuming once things gets more complex. Its easy to make headway when things are pristine and simple. Like a prototype like this. Also as a side note. Google is currently giving away their usage of their LLM API for free. Im not sure how long it will last. They have some decent models. https://aistudio.google.com/ The flash 2.0 pro or Flash 2.0 thinking Models might be a decent alternative to the expensive OpenAIs O1.

Expand full comment