Hi guys! I have documentation full of scanned documents. What is the best way to parse them — should I use OCR or try to convert them into markdown? If OCR is the right approach, which OCR tool would you recommend? I need this in order to search thought data of this documents, these are invoices, contracts, etc. Thank you!

Not a lot of people know about this yet. Google quietly launched Stitch, and it’s one of the most powerful free AI tools available right now. With Stitch, you can create apps, websites, dashboards, or landing pages just by typing what you want. No coding, no design skills. You simply describe your idea, and it gives you a full design, working code, and even a Figma export. Here’s the interesting part: you can actually turn this into a service. You could charge between $500 and $2,000 for rapid prototyping and deliver a working prototype in just a day or two. Freelancers and agencies are already using it to cut their design time by 70% while keeping the same rates. And for local businesses—restaurants, gyms, real estate agents—this is a game changer. They all need apps but can’t afford expensive developers. Now there’s a way to give them something affordable. Most people missed this tool at Google I/O 2025, but the opportunity is wide open right now. You can try it here for free: stitch.withgoogle.com How would you use Stitch to make money or build something new?

Hi! What is best vector database to use when I am working with a lot of document files (around 800k)?

Hi all, I was wondering what is a best way to provide sensitive data to AI agent? I am building a chatbot for SaaS software, where user while previewing the document will have ability to ask some questions about it in chat that is related to it. So, my question is, what is a best way to provide doc content to AI agent? I don't want to store it in vector db, because of potential information leak, neither I want to send document content with each user questions, because it is waste of tokens.

In n8n, I have a problem with a workflow: some node updates are made, but when it runs, it's an old version of the node that's running. It's an active workflow... I tried disabling it while making the change, but that didn't make any difference. I restarted n8n (under Docker), but that didn't change anything either. Do you have any suggestions on what to do to make sure the nodes are properly updated?