The ideas I used to lose

The ideas that get lost

Ideas and thoughts worth remembering come at random moments: while driving, walking, exercising, in the middle of meetings, or talking with someone. Anywhere and anytime when I am not actually working on my computer. For decades, I went through the same pattern: telling myself I would remember the thought and write it down later, which I rarely did. Many useful ideas were lost in the daily hustle.

I am not just talking about practical daily to-do items like errands or bills, but the deeper thoughts that can actually move your life and business forward. Things that get triggered when you are in a headspace that lets your mind wander, or when you are listening to something thought-provoking.

Why most note-taking solutions do not work

Over the years, I have tried many ways to capture ideas worth coming back to: pen and paper, various note-taking apps on my phone and computer, and so on. Each of them worked for a few days and then did not.

Why? There was simply too much friction in the system, and none of them was easily available at the very moment it was needed.

Typing on a computer or a phone is another thing. I type so slowly that I tend to lose my train of thought while doing it. And you cannot do it safely while driving or walking anyway.

Choosing folders, tags, or project names can interrupt the thought.

Why AI may not always help

During the last couple of years, I have tried and built a few ways to manage this with AI tools.

Most did not work. Why?

• They required typing (I am too slow, and I get frustrated and distracted).
• They were too complex technically, including connecting several apps through linking services like Zapier, Make, or n8n. I could make these work, but the solution turned out to be fragile and broke easily as time went by and something changed.
• Notes were either raw, or the summaries somehow missed the point.
• The system could not handle repeated thoughts or connect ideas worth connecting.
• The ideas and notes were not easy to search and recover when needed.
• Some apps simply turned out to be unreliable and not well maintained.

What works: my requirement spec

This finally clicked when I was honest with myself and wrote down proper criteria:

• It has to work by freeform voice: long-form recording that lets me ramble or rant for minutes in a single take.
• It has to work with whatever I always carry. For me, that is an Apple Watch. I often walk or run without the phone, and using one while driving is not safe.
• It has to save the original recordings AND create summaries that I can easily correct or modify.
• It has to save those summaries in a place I control, not behind an app or subscription that may go away.
• It has to be easily searchable both by me and by my AI tool of choice. I do not want to be locked into a single AI tool, as I do not know what my preferred tool will be in a year.
• It has to be searchable by meaning and by simple keyword. This may not matter in the beginning, but in the longer term I believe it will make the whole thing much more useful.

My idea library

After a few unsuccessful attempts, I ended up building an idea library for business ideas and content ideas. I can look at these separately, but one thought can be useful for both. I want to keep that connection instead of making two disconnected copies.

I record my thoughts from my watch or phone. My recording app produces a transcript, and the library receives that text, saves it, and creates a summary. The original transcript is saved alongside the summary, so I can always check what I actually said and compare it with the AI-made summary.

That distinction is important to me. My original words and the AI’s interpretation are different things. I want to know which one I am looking at.

I have a web page where I can look at the recordings, summaries, ideas, and their connections. I can also review suggested connections and accept or reject them. Through a connection to the library, my AI assistant can search the saved material and return the original passages. I can ask a question in my own words, get a refined answer, and ask it to show me the source.

I also specified a separate backup of the library in files I can download and open in whichever tool I want. So I depend on this particular solution to record and process new thoughts, but the thoughts I have already collected stay safe.

Finding something I can use

When I was looking for a subject for this article, I asked my assistant to check the idea library and suggest something worth using for the next newsletter.

It brought back several things I had recorded earlier. These included my thoughts about the advantage AI might give small companies, about learning by building an assistant yourself, and about this idea library.

The small-company subject was the most developed idea in the collection. After considering the options, I chose the idea library because I thought it would be practical for my readers. I was able to go back to the original thoughts and use them as material for the article.

I still had to choose the subject and decide what I wanted to say, but having the earlier material available helped me do that. This is the kind of help I wanted when I started the idea library.

What I would start with

I see so much advice in social channels about prompts, templates, fully defined systems, apps, integrations, and so on. They are often pushed as “the best AI system for…” or with annoying pitches like “Steal my system for…”, as if there were a single way to do this, or a way that is best for everyone. Many of these prescriptive details can result in you building something you do not understand yourself (which means you cannot fix it when it breaks) or, in the worst case, injecting unwanted code into your computer.

Instead of fully designed solutions or building templates, I think a better starting point for most people is an honest brainstorm about what you actually need, what “frictionless” means for you, which tools you prefer, how much DIY tinkering you enjoy, and what your budget is in money and time.

My own experiments helped me understand what I actually needed. Someone else might prefer a simpler system, or be fine adapting something that already exists. The key is to be clear about what you want it to do.

I would also look carefully at the cost and at what I would have to maintain. My earlier trials taught me that I lose interest when keeping the system working becomes another project. Start with what you need most, and add features when you actually need them.

Why a personal idea library

As I grow older, I have felt that while my raw processing power and multitasking ability are decreasing, my capability for synthesizing and describing ideas keeps getting better. Arthur C. Brooks describes this change well in his book “From Strength to Strength”. I highly recommend it to anyone over 50.

A personal idea library simply helps me make the most of my ideas. I have tried to do this in different ways over the years, and I now have something that works for me. I can save my thoughts as they come, return to them later, and decide whether I want to do something with them.

This article first appeared in my newsletter. You can subscribe there.