Evolve with AI
10:29 Share

Using AI to lead our own evolution I think this is a particularly good direction Then yesterday I upgraded my own AI coach logic You can actually take a look at me It's probably last month or maybe April or March There is a article about using the fish's AI as me

It's like using the 627 plug-in to realize an AI coach. The logic is that I put some of my own, for example, Gallup, some of my own advantages and weaknesses, some of my own processing and summary, and then put it in a memory module of Monika within 1500 words.

And then pretend that as long as I tell him about the AI Uncle, he will take all the memories from this part and use a kind of understanding, even with some critical perspective to let me get some insight. For example, after setting up like this, for example, I told him that today I was so stressed that I couldn't sleep well. Then he would say, oh, you may be because there are too many things recently, blah blah blah. You can decompress the pressure like this.

This kind of communication logic is actually a bit like a coach. A life coach, you can talk to him without reservation. He will feel very empathetic about why you are like this. Because he knows what kind of person I am, right? And then he will give me some suggestions. In this way, he will guide me to grow.

I think this thing is very difficult for many users or friends to use because you may need to buy a member of Monica. It costs $100 or $200 a year, so the price is not very cheap. Then it also has some restrictions. The biggest restriction is that its memory module can only be within 1500 words. It will definitely lose a lot of details. The day before yesterday, when I was chatting with my friends, I suddenly thought of this.

Then I saw a friend of mine yesterday, Chen Feng, who also has a similar ability. This ability allows many people, such as Musk and Jobs, to talk to you. You say something to him, and he responds to you from several angles. This is also a good way to play.

I saw her idea and I remembered that I had just talked to a friend about Monica's shortcomings. So I upgraded it. I used my IDE tree to do something similar.

This ability is first of all, I used several files in QI. The first file can write who I am, right? Just to sum up, like I said, the memory in the sim card. Then the second module is how I need it to respond to me. Then the third one, I added a module, which is

What is my most ambitious goal? I hope this goal will keep reminding me that I have done everything I can to be relevant to this matter. As the saying goes, never forget your original intention. This person will definitely remind you. This AI will remind you to never forget your original intention. And then there is a notebook. There is a file that contains the summary of the content I talked to him about every day. For example, after I finished talking to him about this matter today,

I'm going to get a live broadcast tonight. After I finish talking, I might say, "What do you feel after I left?" I will talk to him. After I finish talking, I will let him summarize. In fact, he can know what I am today. After summarizing, it is probably because of this matter, what kind of fluctuations have occurred, how to deal with it, what can I do next,

These things will be stored in a notebook file In the AI IDE tree OK Then the more interesting thing came I will write a smart body I asked him to read the first step in detail

What kind of person I am. The second step is to read each of my diaries. The third step is to know what my ultimate goal is. What is my ultimate goal? What is my ultimate goal? I have to try to guide myself to say that what I do now is related to this goal. The fourth step is to respond to me with a set of requirements. I also added an MCP. I added a Flow Mode MCP.

After I talk to him every day I just want to tell him You help me summarize what I talked about today And help me synchronize to Flowmo He will put that summary note Save it locally and synchronize it to Flowmo You see, it's very convenient OK, so in here I tried it yesterday Oh, it's really, really good That's like

I talked to him once yesterday, right? I summed up the file. Then the second time I talked to him, I said, "Oh, what's going on recently?" "I'm still under a lot of pressure." I didn't say anything, but he read it. After reading it, he said, "Oh, you have to do three things at the same time." "These three things are really stressful." "How should you do it?" "You can distribute it like this." Hey, look, he said he understood me. Because he knows what I've been doing lately. Right? So this is really very, very fun. Very useful.

So I think this is a better logic than the past use of Monica to achieve because Monica she actually does not have a very clear memory strategy she needs you to add that dialogue to add to her memory connection point and that

It's relatively simple. Like this one, you can actually be very specific. You can describe some things. As long as it doesn't exceed the whole model's top and bottom lines. Then the model top and bottom line cloud 3.7 is 200k. 200k is about 200,000 tokens. That might be 300,000 to 400,000 words. But I don't know if there will be a stage. It won't let you pull this up and down the window for a long time. So I don't know if there are 100,000 words. 100,000 words are also

It's enough for me. Last year, I had 50,000 FOMO records. So theoretically, it's enough for me to have a lot of users. But this is actually very similar to "Evolve with AI". How do you understand it? I'm constantly communicating and improving with AI.

Because, for example, when I have a chance and need to make a decision, I can actually ask him. He can understand my perspective from there, and give me some suggestions from a very strong perspective of insight, and give me a long-term goal. So in this way, you can see that he is a bit like an AI mirror. AI makes an AI mirror, and inside the mirror is actually the most perfect self.

When I'm looking at the mirror, I'm interacting with it, and it's actually pulling me towards it. Right? Look, does this look like an "involve with AI"? It's using AI to make yourself more evolved. I think this is still very interesting. Then you can try it out. I recently wrote a course to describe how to use it in detail. Anyway, it feels pretty cool to use it.

That's all for today. What I want to talk about is how to make AI better to help itself evolve. Look, now after Cloud 3.7, there are many models that have the ability to push. And its online window is also very long. So, theoretically, I can give him all the content I chatted with him in the past. Then he can give me some suggestions based on what I talked to him in the past, including my understanding.

Right? This suggestion is based on a very long upper limit and a very strong reasoning ability. Including I set a goal for him. He came together to

to let me evolve. When I talk to him, I will constantly have new ideas. Then I will do actions. Then PDC, right? PD, Plan, Do, Check. I can check with AI. So this will be very helpful to me. I will evolve with AI. Because it

can evolve from 3.7 to 4.0, right? Evolving to 4.5, or even 5. It's evolving, so am I. Then, the way I use it, or the way I cooperate with it, is also evolving. This is not possible with agent. This is not possible with fully automated. Because in the end, plan and do, plan is what you decide together. Because you are ultimately a human being with subjective ability.

And then the degree is definitely done by yourself. He can't help you in the real world. Control you. Right? You still have an objective individual. So this is pretty interesting. Then you can try it. I will write a special tutorial later. If you think

If you are interested in this topic, please leave a comment. If there are more comments, I will write it as soon as possible. Because for me, this is not a hot topic. The more I write, the better. I can use it to use more cases. But I think from the article I wrote in March to now, the core of his thinking logic has not changed. It's still this set.

只不过我加了就用多了会有些进化 比如说记忆 我跟他聊天的笔记内容 我可以让他在本地里面读一遍 再给我回

Then I added a super goal, which is a super target module. I have a ultimate goal. I hope all my actions at the moment can support and help the long-term goal. If not, you have to remind me that this is not to forget the original intention. I think this is still quite interesting. You can try it. Okay, I can try the content I'm talking about today.

Then let AI sort it out I'll see if I can publish an article This will be faster Okay, that's it. Bye