Recently, I've been learning how to use AI to automate work. I installed Xiaolongxia (Little Lobster), but I’m still figuring out how to use it. My goal is pretty clear though. I used to have a Discord server called 'Get Rich Street,' but I lost access to my Google Authenticator and couldn’t log in for a few months while living in mainland China, so I stopped maintaining it. Now it seems like I can use AI to keep it running again. For example, I could create separate channels for the cyclical stocks I follow, keep daily updates on news and industry analysis, and refresh the group’s stock info and thought processes. Other than working out every day, I spend most of my time reading research reports, checking financial news, and scrolling on X. Right now, I manually throw a lot of investment research into a few AI platforms I subscribe to and interact with them to get results. This year, I feel like AI has gotten a bit smarter. The conclusions it gives me are way better than last year, and overall, it’s helped me avoid the big pullback in 2026 (group members know I’ve been saying since January that last year’s gains were about done, so it was time to go light on positions. Later, after the Iran incident, the chemical and power sectors I bought at the bottom did pretty well too). I feel like AI has boosted my 60/100 investment skills to around 75-80/100. But I’ve realized there’s a very important trick to using AI: it’ll follow your lead, confuse you, and even flatter you. You have to constantly remind yourself to use multiple sources, multiple AIs for cross-checking, and repeat prompts to tell it not to sugarcoat things. Otherwise, if you throw out an idea, it’ll hype you up like you’re Warren Buffett. Another key point is to keep the context clean and only discuss one topic at a time.