Day 27 of #buildinpublic Enum - an AI tool for everyone (load multiple data sources and chat).
Tried finish implementing plugin for Crisp, failed. Their webhooks stopped working, even for the experimental project that worked before. Waited for support.
Day 25 of #buildinpublic ChatGPT project (feed GPT your data and query it).
Moved long operations like reading PDF files and scraping websites to separate threads, as a result, had to refactor working with the database ๐ฌ๐
Day 23 of #buildinpublic ChatGPT project (feed GPT your data and query it).
Worked on the problem of indexing big (and even not too big PDF files). Somehow, managed to improve the speed of extracting text from it. Also, fixed the Qdrand client timeout problem. โบ๏ธ
Day 22 of #buildinpublic ChatGPT project (feed GPT your data and query it).
Fixed small and big bugs and found that extracting text from PDF files was not accurate ๐found another library, PdfMiner, and worked pretty nicely but the speed... is slow.
Day 21 of #buildinpublic ChatGPT project (feed GPT your data and query it).
Implemented calculating and visualization limits and usage, also, finished the deployment of the app to AWS, so the product is almost ready for a quiet launch ๐
Day 20 of #buildinpublic ChatGPT project (feed GPT your data and query it).
Added "pricing" and started working on the limits. When you have a business with paid API like OpenAI, giving free plans could be expensive so I don't want to waste more than I can afford ๐ฌ