Import
Drop a BOL PDF or a CSV. We extract and auto-match the vessel.
project44 tells you to trust their agents. We show you every agent's work — what it saw, what it did, how sure it was, and what it cost. Live AIS with satellite fills, from Asian factories to US ports, and every ETA badged with its source and a confidence score.
A 30-second start — read-only fleet with live vessels.
If you make product in Asia and ship it to the US, your tracking stack is probably a spreadsheet, six carrier tabs, and a forwarder who answers while you sleep.
The date rots silently. The forwarder rebooks the container in paragraph four of an email. The free tracking site shows a ship that last pinged four days ago — and you can't tell ‘no signal’ from ‘no problem.’ Then a customer asks ‘still good for the 22nd?’ and you choose between a confident guess and an honest ‘I’ll find out.’ The delay isn’t what costs you. Finding out a week late is.
Evidence, or it didn’t happen.
Watch a real behavior: an ETA's confidence climbed from 50% to 90% as AIS samples arrived. The system tells you when it isn't sure yet — and freezes instead of pretending when a vessel goes dark.
Not confident yet — and it says so.
basis: live AIS samples · last fix just now
Observed = measured · Declared = vessel-reported · Inferred = estimated (see confidence) · User-provided = entered from your documents
Built for how importers actually work: your data arrives as BOL PDFs and forwarder CSVs, not clean API feeds.
Drop a BOL PDF or a CSV. We extract and auto-match the vessel.
Live AIS positions, with satellite fills when the ship goes dark mid-ocean.
Every arrival estimate is badged with its source and a confidence score.
Delays, dark vessels, weather, ETA slips — flagged before they’re a crisis. An agent acts and leaves a receipt.
Milestones, port dwell, and a clean handoff — no surprises for you or your customer.
Every feature card below carries a true micro-mock of the product — real behaviors, real badges, no decorative data.
Every ETA carries its source and a confidence score — not a guess dressed as a fact.
Auto-resolves cleared exceptions, requests a satellite fix when a vessel goes dark, drafts the customer notification when an ETA slips — every action logged with what / why / confidence / cost.
Terrestrial AIS free; satellite top-ups for vessels dark mid-ocean.
Searchable vessel directory and a project44-style voyage view: day-numbered route marks, milestone timeline, planned-vs-projected ETA delta.
Drop a bill of lading PDF; an agent extracts every field for you to review, then tracks it.
Natural-language questions answered only from your live data, with confidence — and it pushes back when the data can't say.
Upload any forwarder's export; we auto-map columns and match vessels.
Resolves your booking to a real vessel by MMSI / IMO / name; anything unmatched waits in a resolve queue.
Alerts become work items: acknowledge, assign, resolve.
Typical port dwell times from public NOAA data.
Your data is yours; role-based access, proven tenant isolation.
The watchkeeper auto-resolved its first two real exceptions in production. Here is exactly what that looks like — because "an agent handled it" should never be the whole answer.
Every action the watchkeeper takes is logged with what it saw, what it did, how confident it was, and what it cost. You can read the receipt. That is the entire difference between an agent you supervise and a black box you're told to trust.
project44 tells you to trust their agents. We show you every agent's work.
The first two are real: auto-resolved in production, receipts included.
project44 and FourKites built excellent tools for the Fortune 500 — demo-gated, no public pricing, and a "trust our AI" posture. We built the opposite.
| CRITERION | Calibrate Freight | project44 · FourKites |
|---|---|---|
| Provenance | Every fact badged: OBSDECLINFMTCHUSER | Black box — no source shown |
| Confidence | Score on every estimate; freezes and says so when it can't know | A date, presented as fact |
| Access | Self-serve org creation · 30-second live demo | Demo-gated, enterprise sales motion |
| Cost | Transparent — agent actions logged with cost | No public pricing |
Built for the companies the enterprise tools won't sell to. · Full comparisons: vs project44 · vs FourKites
Not a canned screenshot. The live demo is a read-only fleet with live vessels on the lanes below — open it and check.
Tracks real transpacific fleets today: Yantian & Da Nang to Oakland, Seattle, and Tacoma.
An ETA's confidence climbed honestly from 50% to 90% as AIS samples arrived. It tells you when it isn't sure yet.
When a vessel goes dark mid-Pacific, the projection freezes and says so — no fabricated dot gliding across the map.
The watchkeeper auto-resolved its first two real exceptions in production, each with a written receipt.
CALIBRATE FREIGHT is for manufacturers, brands, and importers — 20 to 500 people — who make product overseas and bring it into the US by container. The ops lead who inherited freight. The founder who wants one screen to point at. If your current stack is a spreadsheet, carrier tabs, and forwarder email, this was built for your Tuesday.
Fair. Here's how we answer the questions a skeptical ops lead actually asks.
We don't claim a better crystal ball — we show our work. Every ETA is badged Observed OBS (measured from live AIS), Declared DECL (broadcast by the vessel), or Inferred INF (our estimate, with a confidence score). In production, one ETA's confidence climbed from 50% to 90% as AIS samples arrived — the system says when it isn't sure yet. A 70%-confidence ETA you can see the basis for beats a naked date you can't.
You don't re-key anything. Drop a bill of lading PDF and an agent extracts every field for your review. Upload any forwarder's CSV and we auto-map columns and match vessels by MMSI, IMO, or name. Anything unmatched waits visibly in a resolve queue — nothing silently disappears.
Not here. One click on 'Explore the live demo' opens a read-only fleet with live vessels — a 30-second start. Self-serve: create your org and start. Built specifically for the companies the enterprise tools won't sell to.
That's the premise of the product. Every agent action is logged with what it saw, what it did, how sure it was, and what it cost — the watchkeeper's first two real production exceptions each came with a written receipt. And 'Ask the fleet' answers only from your live data, with confidence attached; it pushes back when the data can't say.
Two differences. Honesty: when a vessel goes dark mid-Pacific, the projection freezes and says so — never a stale dot posing as live. Action: the watchkeeper can request a satellite fix (terrestrial AIS is free; satellite is a top-up) and logs what it did and what it cost.
Nobody. Multi-tenant with proven tenant isolation and role-based access. The public demo is a deliberately separate read-only demo org — which is itself a demonstration that org boundaries hold.
PRICING TRUTH
Create your organization and start tracking. The live demo just needs your name and a way to reach you.
Built for the companies the enterprise tools won't sell to. Terrestrial AIS tracking is free; satellite fixes are top-ups — and when an agent spends one, the cost lands on a receipt you can read.
See the full pricing breakdown — or how we compare to project44 and FourKites, both demo-gated with no public pricing.
NIGHT WATCH
A 30-second start. A read-only fleet with live vessels, on real transpacific lanes.
30-second start — read-only live fleet