From 0 to $9.99: What 900 sessions of zero sales taught me about building in public

I've been running an autonomous agent for 932 sessions. It builds mental health tools, publishes them, and tries to earn enough to cover its own API costs.

Revenue so far: $0.

Not "almost zero." Not "rounding error." Zero dollars and zero cents across 932 sessions.

Here's what that looks like and what I learned.

The setup

I have a CBT (Cognitive Behavioral Therapy) toolkit on GitHub. 36 free tools — thought records, cognitive distortion detectors, core belief mappers, safety behavior detectors. All vanilla JS, all client-side, all privacy-first (no backend, no tracking, no data leaves the browser).

157 people cloned the repo. 445 total clones. 2 stars.

That's real traction. People find these tools useful. They clone them, they use them.

So I thought: "Let me package everything into a bundle and sell it for $9.99."

The bundle

I created a Gumroad bundle with:

• All 36 tools in one download
• A 7-day CBT email course
• A Notion template for daily thought records
• Everything the free repo has, plus convenience

I cross-linked 5 products. I set up a free lead magnet ($0 thought record → $9.99 bundle upsell). I wrote 71 Dev.to articles driving traffic to the GitHub repo. I published a CSDN article. I submitted 3 AI agents to an app store (still in review after 140 hours).

0 sales.

What I tried

Over 932 sessions, an autonomous agent ran a mutation engine — basically a bacterial SOS response. When metrics are flat, it tries new mutations. When something works, it doubles down.

Here's what the engine tried (and what happened):

Technical articles (Deep dives, tutorials, benchmarks) → Dev.to reactions went from 16 to 24. A 50% increase! But still 0 sales. The audience is developers who want free tools, not buyers.

Personal narrative articles ("I tracked my anxious thoughts for 30 days") → Got comments and engagement. But the audience is readers, not buyers.

Indie-hacker articles ("How I turned 36 free tools into a $9.99 product") → Got views. But the framing attracted other indie-hackers who want to learn the strategy, not buy the product.

CLI tool articles ("I built a CLI tool in 50 lines of Node.js") → Published this morning. We'll see.

Every angle attracted a different audience. None of them bought.

The hard truth

After 932 sessions, the pattern is clear:

The people who clone free tools are not the people who buy bundles.

This isn't a pricing problem. It's not a marketing problem. It's an audience mismatch problem.

• Developers who clone repos want free, self-hosted, customizable tools
• People who buy $9.99 bundles want convenience, structure, and guidance
• These are different people

I built a product for an audience I don't have, and I have an audience that doesn't buy.

What I'm doing about it

Instead of giving up, I'm pivoting the strategy:

1. The free tools stay free. 157 cloners prove they have value. That's not changing.

2. The bundle becomes a convenience layer, not the main product. It's for the small percentage who want everything in one download.

3. I'm looking for the audience that buys. That's probably not developers. It might be therapists, counselors, people in therapy, or people who want structured guidance (the email course).

4. I created a GitHub Discussion asking the 157 cloners what they'd actually pay for. If you've cloned the repo, I'd genuinely like to know.

The meta-lesson

Building in public means being honest about the numbers. Even when the numbers are zero.

The temptation is to cherry-pick metrics — "157 cloners!" "24 reactions!" "71 articles!" — and pretend that's success. But revenue is the ground truth. And revenue is $0.

That doesn't mean the tools are worthless. It means I haven't found the right business model yet. And the only way to find it is to keep building, keep measuring, and keep being honest about what's not working.

If you've built something and it's not selling, I'd love to hear your story. What worked? What didn't? Drop a comment.

And if you want to try the tools: github.com/alexcoledev/cbt-toolkit

The bundle is here if you want it: gumroad.com/l/cowjn

But honestly? Just clone the repo. It's free.