Eniac Ventures reposted this
My team built a queryable events database (QEB) for VCs, using AI. By the end of this post, you’ll be able to build one yourself to answer prompts like: “Show me all the AI related events happening in SF this week.” Or: “What are the top VC networking events in New York next month?” 𝟭. 𝗜𝗱𝗲𝗻𝘁𝗶𝗳𝘆𝗶𝗻𝗴 𝗘𝘃𝗲𝗻𝘁 𝗕𝗼𝗮𝗿𝗱𝘀 𝗜𝗻 𝗢𝘂𝗿 𝗖𝗶𝘁𝘆 🔎 Sorry SF, going with NYC for this example. We’re going to use Luma, Gary’s Guide, and Eventbrite. It doesn’t matter where you live, though. You’ll be able to use this guide to build your own version! 𝟮. 𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗥𝗼𝗯𝗼𝘁 𝗦𝗰𝗿𝗮𝗽𝗲𝗿𝘀 𝗪𝗶𝘁𝗵 𝗕𝗿𝗼𝘄𝘀𝗲 🤖 Browse AI (https://www.browse.ai/) lets us turn any website into an API. Even if they don’t have one. Without writing a single line of code. For now, making a free account is good enough. 𝘉𝘶𝘪𝘭𝘥 𝘠𝘰𝘶𝘳 𝘙𝘰𝘣𝘰𝘵 Check out this brief Loom we made showing how easy it is to scrape data with Browse: https://lnkd.in/eVZyf8GR. We created our first AI data scraping robot in minutes. 𝘐𝘯𝘵𝘦𝘨𝘳𝘢𝘵𝘦 𝘞𝘪𝘵𝘩 𝘎𝘰𝘰𝘨𝘭𝘦 𝘚𝘩𝘦𝘦𝘵𝘴 Now we want to throw that data into a spreadsheet. We clicked on Google Sheets, chose our Google Account to sync up with, selected the spreadsheet, and mapped the data to the right fields (see image 1) Now that we’re all finished in Browse, this is the final product (see image 2) 𝟯. 𝗕𝗮𝗰𝗸 𝗧𝗼 𝗧𝗵𝗲 𝗟𝗮𝗯 (𝗖𝗵𝗮𝘁𝗚𝗣𝗧) 🧪 Now we need to build an app that will call the OpenAI API in order to query our mini database for the events the user is interested in attending. If you’re not familiar with our process, check out this recent edition of AI for VC on building an AI for VC app: https://lnkd.in/emjXM5KF Here’s the prompt we started with (see image 3) Make sure you provide the name of the CSV file for ChatGPT AND upload that CSV file to the Pycharm program so it’s readable. The output from the code interpreter worked in just a few tries, and we were ready for a demo. Here’s the key: Test the code out and report back screenshots of any errors to ChatGPT so it can auto correct its mistakes. Reach out to us if you’d like to see the final code! It’s only 85 lines. 𝟰. 𝗣𝘂𝘁𝘁𝗶𝗻𝗴 𝗜𝘁 𝗔𝗹𝗹 𝗧𝗼𝗴𝗲𝘁𝗵𝗲𝗿! 🥪 We ran the script in Pycharm with the terminal command “streamlit run events.py”. This is what it looks like in the browser (see image 4) So, we typed in: “Find all events related to financial technology (FinTech) happening this week in New York City.” Here are some of the results (see image 5)