Why we built Droptime (and why the pricing is flat)

Every product carries an opinion about its user. Enterprise roasting software's opinion is that you are a data-entry resource for a quality department. Artisan's opinion is that you are a craftsperson at a machine — which is right, but only until the machine cools down.

Droptime's opinion is that you're a professional running a whole roastery with too few hands, and that what you're missing isn't more data. It's a colleague who's already read the data.

The colleague test

Here's the standard we hold every feature to. Imagine a sharp head roaster who's been with you for years, has perfect recall of every batch, and watches your inventory like it's their money. What would they say to you?

Not: DTR: 18.2%. RoR[480s]: 0.4°F/min. Δ vs. profile: -2.1.

More like: "Your rate of rise crashed to near-zero at 8:10 — a stall right after dry-end. Third time this month on this coffee. Try a smaller gas cut thirty seconds before dry-end instead of at it, and cup it against Tuesday's batch before you change anything else."

Both statements come from the same numbers. One of them requires you to be the analyst. The other one is the analysis. That difference — plain language, comparison against your own history, one or two concrete next steps — is the entire product thesis. It's why the tagline is what it is: software that talks to you like a colleague instead of a lab instrument.

Grounded, or it doesn't ship

The obvious objection to "AI roasting assistant" is the right one: language models make things up, and a roaster will catch a fabricated first-crack time in about four seconds — and never trust the tool again. Rightly.

So Droptime's AI is built on one non-negotiable rule: it may only cite numbers that exist in your database. When you ask "why did my RoR crash at 8 minutes?", the model doesn't answer from vibes — it queries your actual curve arrays, your event markers, your prior-batch statistics, and reasons over what comes back. When a cupping transcript is ambiguous about whether you said 7 or 7.5, the field is left blank for your review, not filled with a guess. When you ask about data you haven't entered, it tells you what's missing instead of improvising.

This is more work to build than a chatbot. It's also the only version of this product that deserves to exist.

Why the pricing is flat (and staying that way)

The most-resented thing about the incumbent isn't a missing feature. It's the pricing model: per-machine volume tiers, where your software bill climbs with your production. Think about the incentives that encodes. The vendor profits from your growth without contributing to it, and every capacity milestone you hit comes with a tax.

We priced Droptime as a rejection of that model, and we mean it as a positioning statement, not a discount:

• Hobbyist — free. Roast logging, curve analysis, Artisan import, twenty AI analyses a month. Free forever, because the person roasting on a converted popcorn machine this year is the roastery owner in three.
• Roastery — $79 a month. Everything: inventory, blends, planning, cupping, chat. Whether you roast five thousand pounds or a hundred and fifty thousand.
• Roastery Pro — $149 a month. More AI actions, a team, wholesale sync, an API.

No volume tiers. No per-machine fees. And the pledge that backs it up: your data is yours, export anytime. One click produces everything — structured roasts, your original .alog files, the inventory ledger, blends and their versions, cupping forms and audio — in an archive that re-imports losslessly. An export you could leave with is the only honest proof that we expect you to stay because the product is good.

Where it goes from here

We built the import-first version deliberately: Artisan and CSV in, intelligence out, no hardware between you and value on day one. Live probe capture is coming, and it will land on the same data model your imported roasts already use.

(Update, July 2026: it came — as a free, open-source desktop app. Meet the Droptime Logger.)

But the order of operations is the message. We didn't start with cables and dashboards. We started with the conversation you'd want to have at the cooling tray — what happened, what's running out, what do I roast tomorrow — and built the software that can hold up its end of it.

If you've got a folder of .alog files, you're two minutes from your first readout. We'd love to hear what your batches have been trying to tell you.