Almanak has recently launched a new product and has conducted some research before, so I would like to take advantage of the popularity to talk about this project. Their main slogan is: One person runs one on chain hedge fund. I checked the word Almanak and it means' yearbook '. Generally, books like to use this name, such as' Chinese Cultural Relics Yearbook'. However, what this project does is AI agent+strategy, with various strategies flying around and added together, it can also be called a fund strategy yearbook. So, I tested their product and the input command was: Make mainstream coins (WBTC, ETH) and stablecoins (USDT, USDC) as LPs on Uniswap. 1. Which pair has the highest APR and on which chain? This may involve several possibilities for 2 * 2 * chains. 2. Determine the price range that should be set through short-term and medium-term judgment 3. Calculate the proportion of principal that should be invested in mainstream coins and stablecoins based on the price range (assuming each position is invested in $100K) Then Almanak's AI asked a few questions and began to capture and analyze. Finally, it was analyzed that the ETH/USDC liquidity on Base had the highest APR among the above options, and the range of 1600-3300 was chosen. Next, we will directly enter the development, testing, and deployment process. PS: I really don't treat AI as a human. The entire process only takes 10 minutes. Next, if ETH is really injected into it, the wallet will start executing the strategy directly. Speaking of which, from the perspective of fund risk control, large funds should not be directly placed in AI written code. If large funds are really used, it may still be necessary to introduce real people to proofread and find professional code auditing agencies to audit before they can be used. The code generated by Almanak is suitable for small funds and testing ideas, also known as "idea validation". For example, at the end of last year, an institution applied for a night market Bitcoin ETF, established a position in Bitcoin after the US stock market closed, and sold it at dawn during the day. At first glance, this idea sounds good to you, but you still need to run a real test to see how it actually works. You throw in $100 and directly tell Almanak to run the Vibe Coding code for ten, five, and a half months of actual testing. You will naturally have a result in your mind, saving a lot of time in an instant. That's also why programming with a mouth is called Vibe Coding, because it's a feeling (Vibe), driven by feeling. Speaking of yearbooks, since they are all strategies developed by Vibe Coding, there is also a strategy square where various strategies are available. For example, running on Pancake, BTC-BNB strategy, and AI scoring, evaluating security and yield, and so on. If you are interested, you can come and test them all. Each ordinary user can use three strategies (Signal) for free, but if there are too many strategies written, they will need to recharge as a member. The link is in the comment section.