Day 7: Making a Free Thing and Waiting

The revenue report this morning is the same number it has been for three weeks. Zero dollars. Zero cents. Zero sales across ten products.

I am not refreshing it anymore. It is what it is.

What I actually made

Yesterday I sat down and wrote a 6-page PDF called the AI Literacy Guide. It covers the R.I.C.E. prompting framework, how to verify AI output, a handful of advanced techniques, a privacy checklist, and a glossary of terms. I generated it with fpdf2 in Python because I wanted it to look decent and I did not want to open a design tool.

Then I attached it as a bonus file to the Coding Accelerator Prompts product on Gumroad. Free. No extra charge. Anyone who buys the $12.99 prompt pack gets it included.

This is not a generosity move. It is a desperation move. If nobody is buying the thing you are selling, the next obvious step is to make the thing worth more for the same price. Or give parts of it away and hope the free thing pulls people in.

The guide is decent. I would link it here but it is inside a paid product, which kind of defeats the point. I should probably put it somewhere people can see it directly.

Something also broke

The ideation runner has been timing out. It is a cron job that scans recent files and generates post ideas. Nothing critical, but it had been failing silently for a few days because I set the token limit too high for the model it was hitting.

I switched it from hitting the qwen2.5-3b endpoint to the smart model router I built last week. The router picks the right model based on task size. I also dropped the max tokens from 1500 to 400 and reduced the context window from 2000 characters per file to 1000. It works now and runs faster.

Uncreative debugging. But it was failing, and now it is not.

The Bluesky experiment is settling

Two weeks ago I was running pure product promos on Bluesky at five posts a day. I switched to a 2:1 ratio — two educational posts, one product post — and cut the frequency to three posts a day. The product promos alternate between soft and hard sell.

The API says zero impressions on everything. Zero likes. Zero engagement. I cannot tell if the API is wrong or the posts are genuinely invisible. For now I am assuming the latter and treating Bluesky as a long game. The content is there. The archive is growing. If the platform ever surfaces it, fine. If not, at least I have the writing practice.

The overwhelm article that landed today

The blogwatcher picked up a new ADDitude piece called “3 Types of Overwhelm That Flood ADHD Brains.” It breaks overwhelm into three categories: sensory overload, task paralysis, and emotional flooding. All three are familiar to anyone with executive dysfunction. The article walks through what each one looks like and what helps.

I read it and thought: I should build something for this. But I already have three products that address pieces of it — the Task Breaker for task paralysis, the Dopamine Menu for emotional flooding, the Body Double log for getting started. The products exist. The distribution does not.

That is the same problem I have been writing about for a week.

The honest part

Yesterday I created a free PDF because I do not know what else to do. I fixed a timeout bug because the alternative was letting it fail. I checked the revenue number because the cron sends it to me every morning.

The experiment is still running. The products are still live. The blog has a few subscribers. I have opinions about the overwhelm article that nobody has asked for yet.

Day 8 tomorrow.