I'm writing up some of the verticals individually because each one represents a real audience and a real problem that I want to describe on its own terms. First up: cheesemaking. It's live, it's on GitHub, and it's the first of our verticals to feature an AI vision component.
The pitch is simple: home cheesemakers need to log their batches from milk to aged wheel. They need to track temperature and humidity in their aging vault. They need to record tasting notes. And when they're not sure what's happening with a particular wheel — is the rind developing correctly? is that the right mold color? — they want a second opinion that isn't "post to Reddit and wait three days."
cheesemaking is the app for that.
What It Actually Does
Four core things:
- Batch logs. Every wheel you make gets a record: name, cheese type, milk source, start date, aging target, current status. SQLite under the hood — the whole batch history is one file on your disk.
- Aging vault tracking. Daily or as-needed log entries for your aging space: day number, temperature in °F, humidity %, free-form observations. Builds a trend line so you can catch drift before it ruins a wheel.
- Tasting notes. A 1-10 rating plus fields for flavor, texture, and aroma. Over time you build a record that can actually answer "which Brie was the best one?"
- AI vision aging assistant. This is the differentiator. Upload a photo of a wheel. Claude reads the rind color, estimates mold coverage percentage, guesses the aging stage, and suggests tasting-note keywords to watch for on your next cut.
Why the AI Piece Matters
Home cheesemaking is a solo hobby for most people. You don't have a master cheesemonger in the next room to come look at your wheel. You've got books, YouTube, and a few Discord servers. That's fine but it's slow, and by the time you get a second opinion, the wheel has changed.
The vision component isn't replacing expert judgment — it's giving you a fast first pass. "This looks like a Camembert at around day 14-18. The rind is developing white bloom consistent with Penicillium candidum. Watch for ammonia notes; consider cutting soon." That's genuinely useful, and it's the kind of thing that was impossible to ship as a small app until very recently.
The call into Claude goes through our shared dcch wrapper, which means retry logic, model fallback, and template fallback are all handled. If Claude is down, the app still works — you just don't get the vision analysis. If you're self-hosting without an API key, same deal. The vision feature is additive, not load-bearing.
The Build
Flask + SQLite. Single-password auth via our template. Stripe wiring for the hosted tier. The landing page at dangercorn.net/cheesemaking follows the same house style as every other vertical. About 1,200 lines of custom Python on top of the template, not counting the Jinja templates.
To run it yourself:
git clone https://github.com/Dangercorn-Enterprises/cheesemaking.git
cd cheesemaking
pip install -r requirements.txt
echo "AUTH_PASSWORD=pickyourown" >> .env
echo "ANTHROPIC_API_KEY=sk-ant-..." >> .env # optional
python app.pyThe app binds to a deterministic port thanks to our port hashing scheme. Open the URL it prints, log in, start a batch.
Pricing
Three tiers, same as every Dangercorn vertical:
- Self-host: $0 forever. Full features. Bring your own Anthropic key for the vision piece.
- Hosted Pro: $39/mo. Fully managed, unlimited batches, AI vision included (no BYOK), aging-vault temperature alerts, PDF batch reports.
- Cheesemonger: $149/mo. Everything in Pro, plus multi-user / shared vault, inventory and sales tracking, custom label generation.
The hosted tiers are in early access — the pitch is: self-host today, email tim@dangercorn.net to get on the waitlist, we'll flip hosted on when we have enough signups to justify the compute. The self-host tier is the same code, so you're not giving anything up by going that route.
Who It's For
The people I've been talking to about cheesemaking fall into two groups.
Group one is the serious hobbyist. Maybe 5-20 wheels in aging at any time. Has a basement or outbuilding dedicated to the hobby. Knows what a cheese cave is. Uses words like "lactic acid bacteria" and "washed rind" in casual conversation. These people are currently keeping records in Excel or a notebook, and they want something purpose-built.
Group two is the small commercial operator. Not a farmstead creamery — too small for the enterprise cheesemaking software. Maybe 50-100 wheels a season, a weekend farmers market presence, considering scaling up. These people need the batch tracking for record-keeping reasons (and eventually FDA compliance, depending on the state), and they're willing to pay $39-149 a month for software that fits.
Neither group is being served by anything that exists today. There's one enterprise cheesemaking software product. There's an Excel template floating around. That's it.
The market isn't big. Maybe 15,000 serious hobbyists and 3,000 small commercial operators in North America total. That's a $2-3M annual ceiling at 100% saturation. Too small for anyone with a real fundraise to chase. Just right for a vertical we can build on top of our template in a couple of weeks.
What's Next
The biggest feature request so far is label generation — "I need to print a label for every wheel I sell, with the type, batch number, aging date, ingredient list." That's in the pipeline for the Cheesemonger tier. Planning also to add recipe tracking (ingredients, temperatures, times for each cheese type) and moisture loss tracking via periodic weighings over the aging cycle.
If you're a home cheesemaker (or know one), I'd love a few real users on this to tell me what's missing. Self-host it, send feedback. The feature list is still pliable.
And if you're just here for the code: it's public at github.com/Dangercorn-Enterprises/cheesemaking. Same template every other vertical uses, same patterns, nothing mysterious.