Foresight™ — memory, not magic.
The only marine app that learns your boatEvery marine app can tell you an oil change is due in 40 hours. That is arithmetic. theONE is built to tell you something arithmetic cannot: this is starting to look like the thing that went wrong last July.
The problem with a service interval
A service interval knows one fact about your boat: a number went up. It does not know that the last time the starboard temperature crept at cruise, it was a raw-water impeller, and that you found it because the telltale spray looked weak, and that it took a haul-out to prove it. That knowledge exists — it is sitting in your logs — but in every other app it is inert. It goes in and it never comes back out.
Your boat has already told you what is about to go wrong. Nothing has ever been listening.
What Foresight actually does
Every observation, issue, incident and work order you log is a node. When one leads to another — a temperature note that becomes an issue, an issue that a job closes — theONE records the link, not just the two ends. Over time those links form chains, and a chain that reached a conclusion gets that conclusion stamped on it: fixed and verified, worked around, not reproducible, deferred.
The stored shape of a chain is called a signature. It holds the structure of what happened — which kinds of record were involved, how they were connected, who acted, and the conditions at the time: sea state, engine RPM band, time of day, where you were. It does not hold your wording. That is deliberate, and it is the whole trick.
Why it matches shape and not words
If Foresight matched text, it would need you to describe the same fault the same way twice — which nobody does, and which would make it useless across two different engines, two different crew, or two different years. Matching structure means a chain that began on the port main can raise a flag on the starboard one, because what repeats is not the sentence. It is the pattern.
A candidate is only compared against prior chains of the same kind, and only against chains that actually reached an outcome. A guess with no ending teaches nothing. And a match has to score at least 0.7 to say anything at all — below that theONE stays quiet rather than filling your inbox with maybes.
Three engines
| Engine | What it watches |
|---|---|
| Foresight™ | Repeating causal chains. "This looks like the chain that ended in the July alternator failure." |
| Hours Watch | The arithmetic, done properly — intervals coming due, intervals overdue, burn-rate anomalies, hour logs going stale, and hour meters that disagree with each other. |
| Departure Windows | Ranks the next 72 hours on four criteria — daylight, wind at or under 15 knots, no active marine alert at your position, and a favourable tide — so you know when to leave, not just what the weather is. |
Two of them share one inbox: Foresight and Hours Watch both raise into the same list, so a captain does not have to check two screens to find out his boat is trying to tell him something. Departure Windows sits with Ready Now instead — it belongs at the moment you are deciding whether to go, not in a queue you read later.
No black boxes
Every warning Foresight raises carries its evidence on the face of the card: a PREVIOUSLY block naming the outcome it matched, the note somebody wrote when they closed that prior chain, and who closed it. Not a score with no provenance — the actual prior ending, in the actual words it was recorded in.
From there, View trace opens the record inside its own history, with the links in and the links out drawn explicitly. Tap any one of those links and a Why? pane shows how that connection was scored and when it was established. So the chain is walkable, one honest step at a time.
This is not a courtesy. A warning you cannot audit is a warning you will eventually learn to ignore, and an alert a captain ignores is worse than no alert. If Foresight cannot show you why, it has not earned the interruption.
Honest about the edges
It needs history before it says anything. Foresight reasons from your boat's own record, so on a brand-new install it has nothing to reason from and it will be quiet. That is correct behaviour, not a fault — but it means the value builds over a season rather than arriving on day one. Hours Watch and Departure Windows work immediately; the pattern-matching earns its keep as you log.
It only learns from chains you finished. A job closed with no reason teaches the engine nothing, which is why closing a record asks how it was resolved. Answer honestly — including "worked around" and "not reproducible" — and the engine gets sharper. Close everything as "fixed" and you are training it on fiction.
It is a memory, not a prophecy. Foresight tells you what your boat has done before and where the present resembles it. It does not predict failures it has never seen, and it never replaces a competent mechanic or your own judgement.
Why we built it
Because the information was always there. Boats are astonishingly well documented and astonishingly badly remembered — the logs exist, the invoices exist, the notes exist, and they are scattered across a binder, a spreadsheet, a group text and a mechanic's memory. The failure is not data collection. It is that nothing ever reads it back.
Foresight is the part of theONE that reads it back.