Lower the bar with transcripts

I’ve been recording myself talking instead of writing. Then letting AI clean it up afterward.

When I write, I’m constantly stopping. Should this be bold? Does this sentence need a comma? Is this the right word? I’m editing while generating.

When I talk, I can’t format while speaking. I can’t fix spelling. I just follow the thought.

If one idea splits into three, you just keep talking. You don’t feel locked into the structure you started with. Maybe this becomes one post or maybe it becomes three. You don’t have to decide upfront.

With writing you often feel committed to the outline you had in mind. You start with “this is my post about X” and that becomes a constraint. But when you’re talking and X reminds you of Y and Z, you just follow it. Later you can carve it up into separate pieces.

The generation mindset and the evaluation mindset compete for the same mental resources. When you’re filtering for quality while generating, neither process works well.

So I record. I ramble. I follow tangents. Then the AI translates it into something readable. It handles the “should this be bold” phase while I focus on getting ideas out.

The low bar approach captures half-formed thoughts and weird tangents that sometimes lead somewhere unexpected. You can raise the bar later when you switch modes.