knowledging

TIL in the Age of Generative AI: Writing with a Focus on Soft Skills

Summary

• TIL (Today I Learned) seems redundant in the age of generative AI.
  • Traditionally, TIL was used to jot down snippets oriented around hard skills, such as procedures, commands, and configuration values.
  • However, with generative AI, you can simply ask the AI when needed.
• Q: So, is TIL no longer necessary?
  • Ans: No
  • You just need to write with a focus on soft skills.
  • For example, keep logs of your daily work, jot down tasks and ideas, record thoughts and introspections.
  • Provide these to a generative AI to gain insights.

Background

Are You Familiar with TIL?

TIL (Today I Learned) refers to jotting down the knowledge you gained each day in formats like Markdown. Originally a Reddit culture, it also became popular on GitHub for a time.

Here are some examples:

• https://github.com/jbranchaud/til
• https://github.com/thoughtbot/til
• https://github.com/simonw/til
• https://github.com/milooy/TIL

Typically, it involves leaving snippets such as commands, code syntax, and configuration values. On GitHub, you can easily browse and search these via a browser, and of course, read them on a local terminal or IDE. The more snippets you collect, the more convenient it becomes, and if you make them public, it showcases your diligence as an engineer, benefiting your career.

TIL Has Become Unnecessary

Thanks to generative AI, it’s faster to just inquire when needed.

If the knowledge isn’t domain-specific and unshareable outside, TIL is no longer needed. Maintaining TIL in an analog manner without adapting to the generative AI trend appears almost comical. While it’s a personal choice how one maintains knowledge, not adopting a generative AI-first perspective in our AI-driven era seems risky in the long run.

But TIL Remains Beneficial!

Thinking strictly in terms of doing it or not doing it is foolish. There’s a middle ground — a hybrid. In other words, you can continue TIL while leveraging the power of generative AI.

TILS

TILS (Today I Learned Softly) involves conducting TIL with a focus on soft skills.

Traditionally, TIL dealt with hard skills like commands, code, and configuration values. Instead, TILS involves writing about soft skills-related information like your tasks, ideas, and insights. By providing these to a generative AI, you can extract insights.

How to Write TILS

There’s no definitive correct way to create it, but please keep the following concepts in mind.

• Accumulate daily memos.
• Feed accumulated memos to a generative AI to gain insights.

For instance, if doing this with VSCode + Cline, you might start with the following structure:

• https://github.com/stakiran/tils

Citing hereinafter:

This repository stores TILS (Today I Learned Softly).

## About TILS
TILS (Today I Learned Softly) is a system for recording and organizing the soft skills learned daily. Originally, there was an initiative called TIL (Today I Learned) that organized hard skills such as commands, code snippets, and configuration values in Markdown. Please consider this as its soft skills version.

## About the structure of the repository
The directory structure is as follows:

- (project root)
    - daily/
        - 20251122.md
        - 20251123.md
        - ...
    - knowledge/
        - 20251122.md
        - 20251123.md
        - ...
    - agent/
        - taskmanager.md
        - peoplemanager.md
        - ...
    - insight.py
    - today.py

Under the daily/ directory, one file per day is saved, and you can write any daily memo. This includes work notes, memos about tasks and ideas, URLs to read later, brief diaries, etc.

The knowledge/ directory outputs the results generated by the **TILS Agent**. The input comes from the daily/ and knowledge/ directories, and insights corresponding to each day in the daily/ directory are output.

Under the agent/ directory, prompts for the TILS Agent are saved. One file represents one agent. For example, if you were to create an agent like a "Task Manager" that organizes remaining tasks and proposes the next tasks to undertake, you would create a file such as agent/taskmanager.md and write the appropriate prompt.

Additionally, there are several scripts at the root.

- today.py
    - Generates and opens today's memo. If it's already generated, it opens the existing memo.
- insight.py
    - Uses today's memo and the prompts from all TILS Agents as context to query a generation AI.
    - The answers are output to the knowledge/ directory.

What TIL Users Should Keep in Mind When Starting TILS

• Likely, you cannot make it public.
  • Soft skills are often personal and confidential, thus cannot be made public as is.
  • Even if made public, you can only share the insights part generated by the generative AI.
• Think in terms of memo and knowledge.
  • Memos are the context you provide to a generative AI, scribbled by you daily.
  • Knowledge refers to the “insights” or “understanding” derived by feeding the memo to a generative AI.
  • If you wish to implement TILS, you need to establish both the method of writing memos and the system of extracting knowledge from the memos through a generative AI.

Conclusion

In the modern era where generative AI is available, the way we conduct TIL is evolving. The focus should shift towards soft skills. For that, you need a system of daily memos and a process of passing these through a generative AI for insights. This entire concept is named TILS (Today I Learned Softly).

In the age of generative AI, soft skills are more crucial than ever. Since soft skills are highly personalized, like engineering, they can be deepened through writing and contemplation. TILS serves as a valuable tool for that purpose. Until next time.