What you’ll get from this lesson: Create an index for finding a task’s information again, record the methods used, and understand the different layers of memory.
Prerequisites: No software installation required. Prepare a work example that can be shared publicly or has had sensitive information removed.
If I want AI to understand my needs better over time, I first need to leave behind things it can refer to. In the original recording, I used the term “second brain” to describe this accumulation. It includes personal information, the tools I use every day, and the methods I follow when working.
Start with the Information Needed for One Task
Choose one task, such as organizing material for an article you want to write. It may require past notes, verified information, and your preferred article structure. Clearly record where the information is located so that next time, you do not have to rely entirely on recalling it from memory.
At L0, you can start by organizing things in ordinary notes or folders; there is no need to install a database first. The worksheet accompanying this article is only for taking stock of your materials. It does not ask you to modify L1’s SOUL.md or MEMORY.md.
Keep a Record of Your Information, Tools, and Methods
An information index records where files are located, what they are for, and which parts need updating. A tool list describes the capabilities available to you, such as reading files or organizing spreadsheets. A methods record preserves your article format, the order of work, and the points that require human confirmation.
You can put all three types of information on the same worksheet without splitting them into a complex system for now. What matters is being able to pull out the part you need when a similar task comes up again.
Distinguish the Current Conversation, Saved Memory, and History Retrieval
Content that is still visible in the current conversation is different from information saved separately in a file. Within the scope of this course’s L1, the Agent writes specified information to MEMORY.md only when the user explicitly asks it to remember that information. L1’s conversation history is stored only in memory and is cleared on restart; memory files already written to disk are a separate matter.
Later, context management addresses how to continue the current conversation, while history retrieval addresses how to find information from further back. This article only helps you understand the distinction. Saving a worksheet does not automatically implement these features.
A Note Needs to Go Through Saving, Retrieval, Use, and Updating
Take a work note that has had sensitive information removed, and first record its location and date. Then try to find one piece of information that answers your current question. Confirm that it actually appears in the original text before using it. When information becomes outdated, mark the parts that need updating.
For example, if a price in your article materials has changed, you cannot keep using it just because the old note is easy to find. Retrieving information is only the first step; you also need to check whether it still applies.
Today’s Deliverable Is an Index You Can Check
After filling out the worksheet, ask yourself, “The next time I do this task, which source will I look for first?” If your answer is still to throw everything in at once, narrow the scope again.
Start by keeping the information you actually use, then gradually organize it into a second brain you can maintain. In the next article, we will map out how data flows and decide what can be sent to the cloud and what needs more careful handling.
👉 Start L1-01: build your first Agent →
L1 is free after you sign in. Start with L1-01, follow it step by step, and you'll have an Agent of your own.
Not ready to start yet? That's fine. Scroll to the bottom and tap "Next lesson →" to keep reading the free articles, then come back to L1-01 when you're ready.