Optimize for Your 'Brain' Tokens, Not LLM Tokens

The more I play around with Claude Code, OpenClaw etc, the more I feel the need to optimize for my cognitive efficiency. These tools need a lot of context, clear communication and well thought-out, well reasoned instructions.

In 2023, my manually written PRDs were larger than ChatGPT prompts. In 2026, most of my prompts or claude.md instructions are now longer and rich in context just to get a decent output (code, apps, text, media etc) from AI tools.

The more I do these things, the more I realise that I am spending a lot of time just thinking/reasoning. The friction between ideation and execution has come down dramatically.

The 'OpenClaw' situation says it's only a short time before the world comes up with a globally accepted SOP/framework/policy for agent<->agent comms. This might be human readable. So, what matters is do I really know how to clearly communicate what I want. This means I should start thinking well.

To instruct them better, I need to think better. To think better, I need to reduce noise. To reduce noise, I need to remove unnecessary inputs. And to have quality inputs, I keep feeling the need to read a lot of fiction, philosophy and start learning guitar again. And, I am not even a bookworm. All of this makes me want to stop 'inputting' short form content, reels and just junk, meaningless stuff.

I feel that the overall outcome of the AI-driven world is where humans will have the opportunity to become who they really are. Without that clarity, you're just generating junk, with fancier tools.

You might have to start protecting your 'brain' tokens. You are going to need a lot of them. It's way more costly than LLM tokens.

"Who looks outside, dreams; who looks inside, awakes." — Carl Jung

Be bored.