Show HN: Personalized financial literacy book for your kid
cointales.aiBasically, my daughter is 11, and I’ve been trying to stay engaged with her studies at school. We all know that modern schools don’t pay much attention to financial literacy. When I was in school, it was the same. I self-studied economics, and though I wasn’t sure why at the time, it ended up helping me a lot in life. These days, most kids don’t know what inflation is, what compound interest means, what a mortgage is, or how banks work.
I’m a software developer, so I started thinking about how I could help solve this problem, especially with all the AI advancements happening now. I wanted to build something fun and interactive for the modern TikTok generation. Working on this with my daughter, we figured out that the best approach would be to create a small AI-generated e-book or physical book, featuring a child’s unique picture that is relevant to the topic covered, and a one-page summary per topic. Honestly, I don’t think kids can focus on more than one page at a time these days.
I tested dozens of AI image generation tools and believe I’ve found the best one. At least, my daughter loves the product and has learned a lot of new financial concepts along the way. I think I’ve built it to the point where I can show it to the world , and I need more people to actually test it and give me their feedback and opinions , especially on image quality.
Of course, we’d eventually like to make some money from it, especially after spending about a month of my time on it and with each API call costing us a bit. But for now, you can upload your child photo and generate three pages for free and give us your valuable feedback which we’d truly appreciate.
Not even using the SOTA models for the samples page, bit shameless no?
Well, I've tried a few and we didn't like the results, can you elaborate which ones you have in mind because there are many?
Look at the text rendering in the Elizabeth one
Well, you're right I've been struggling with the text rendering, in fact I don't need a text at all but the model keeps generating it despite of the strict instruction in the prompt not to, do you think that DALL-E for example will solve this problem?
SOTA right now is nano-banana 2 and luma uni-1, but a much broader point to be made is why should anyone use this service given its low quality and triviality
Thank you for your honest feedback, I will try to increase the quality as of triviality it is your opinion of course, you may be right though, but I will keep grinding.