Bean On Bar for iPhone and Apple Watch
A cone has no opinion. What decides your cup was decided long before the water hit it — at altitude, at the picking table, at first crack. Bean On Bar reads the bean first, and builds the recipe from that.
On the App Store in September.

01 — The product
Built around the coffee habit loop: identify the bean, understand the context, brew with guidance, log what happened, and make the next cup easier.
Context Scores are automated reads of visible listing detail, not verdicts on quality.

At the counter
This is the whole of it. A bag still in someone’s hand, a phone, and eight lines in view — no booth, no tripod, nobody asked to wait.
The reader says slow down before it offers to capture. A confident wrong read costs more than a second attempt.



02 — The loop
Plenty of tools log a roast, and plenty log a brew. What is unusual here is that the file moves in both directions — your roaster’s .klog comes in, a green-adapted .kpro goes back out, and what the cupping table said becomes the next roast’s edit.
Roast
Bean temperature against rate of rise, read from your own .klog.
Brew
What the scale saw, against what you meant to pour.
Rest
The window this lot is likely to be at its best.
What a roaster chooses to disclose is the whole of what can honestly be scored. Everything the app says about a bag is a read of what is already printed on it.
And everything you do with it is timed on your wrist rather than on a phone propped against the kettle. The watch runs the whole brew — bloom, pours, drawdown — into the same record. Following along is free; starting the brew from the watch itself is part of Pro.
How the watch works03 — Capabilities
The native app is strongest when it stays close to the moment: read the bag, choose a starting recipe, run the timer, taste the cup, keep the record under your control.
Capture or choose a bag photo and let Apple Vision read the label on-device. Every field the app suggests — roaster, bean, origin, variety, process, roast level and date, tasting notes, price, weight — stays yours to correct before anything is scored or saved.
Visible label detail becomes a Context Score with its reasoning shown. Cited reviews, competition recognition, and roaster-stated cues appear as separate Taste Signals, so the app never invents review authority it does not have.
Start from an unknown, identified, or saved bean. Choose your brewer, scale the dose, translate grind ranges to the grinder you actually own, and adjust by taste rather than by recipe card.
Price per gram is compared against your own saved history in your own currency. No invented community average, no leaderboard, no number that needs other people to exist.
Plan a trip, shortlist cafes that survive going offline, log what you drank at the counter, and get told when a bean you loved abroad turns up on a shelf near you.
Saved beans, notes, recipes, and brew logs export as JSON, with optional user-controlled iCloud backup. The notebook stays account-light and movable, and we never hold it. The exception is a cup you deliberately share, which becomes a readable link.
No analytics SDK, no advertising SDK, no tracking identifier, no account. Bean On Bar declares NSPrivacyTracking = false in its privacy manifest — a declaration Apple publishes but does not audit — and your notes, label text, photos, prices and locations never leave the device unless you share a cup on purpose.
A recipe is a record of what someone else’s water and someone else’s grinder once produced. The point was never to follow it exactly — it is to know which number to move when the cup in front of you is not theirs.
What the label is telling you