Hi Guys 🙋‍♂️ Anyone here with https://dify.ai/ expertise? If yes, DM me. Potential business opportunity is on the horizon.

Hey Fellow Makers! 💜 I’m helping a client find new potential B2B customers. Here's what I have: - A CRM with companies they've already worked with (very basic data) - A huge raw unfiltered database of 450,000 Hungarian companies - A basic idea of the ideal client profile (ICP) My goal is to: 1. Use AI to analyze the CRM list and define a precise ICP (based on company attributes like industry, size, etc.) 2. Use this ICP to automatically scan the big rawdatabase, then evaluate and score each company (based on revenue, employees, activity codes, etc. + ideally also web search results) to find new high-potential leads. I've done a similar project using Make+Perplexity last year, but I wonder if anyone has a best practice idea. I don't want to build a custom machine learning system and would prefer not to build this from scratch. 🙋‍♂️ Has anyone tackled something similar? Any tools, use cases, or workflows you’d recommend?

Hi guys, I'm trying to set up a Make workflow that can retrieve company data using HTTP request + AI modules. I want to avoid manual/semi-automated web scraping tools and business intelligence platforms that cost a fortune (if possible). I would use HTTP Get request+OpenAI modules to get content from a specific website (Text parser + JSON), extract relevant information and store them in a datasheet. I want to find a way to use these new company data and search for phone/email address if there is any. My problem is, that in many case, OpenAI and Perplexity search functions don't find these, even if they exist. I also tried using "https://www.google.com/search?q=...." in another HTTP get request. It's not bad, but I wonder if there is a more elegant, sophisticated way to do this. I also experienced that sometimes Google excludes web search via an API call. If anyone has a bes practie, please share your insights. Thanks, Janos

Hi guys, I have this 300+ page book to translate from Hungarian into English. I run a one-page test with GPT4o, and the result was very promising. So in theory, the whole book can be translated page by page via an API call-based automation in Make.com. My problem is: if I split the whole document to pages, it will lose context, so the translation will be distorted (topics and paragraphs overlap each other. Not every page is a new section). Question: dou you think it's possible to do non-manual editing to break a large text into small (~1 page) parts with AI and iterate over them with an LLM module? If yes, how would you approach this?

Hi everyone, I have this project of mine putting together a 3-minute promo video about my AI automation projects. I have 8-10 scenes where I want to use 10-second scenes generated by AI tools from text prompts. Here is an example I want to visualize: "PDF-Processing Bots that analyze business reports, research papers, or transaction records, then update your database, perform complex operations, and notify your team — all automatically." The problem is that most tools I've experimented with so far (Runway, Kling, Pictory) are not capable of adhering to text prompts they are more suitable for animating still images. I use LLM-s to write me script prompts, but I couldn't find any useful tool for this kind of task. Maybe anyone has an idea or tip to share?