Price it sharper. Fix it sooner. Collect what you earned.
CapesizeOS reads your market, prices the voyage on what your vessels actually do, checks both sides of the chain before you fix, and assembles the claim before the bar runs. Your traders decide. The system does everything up to the decision.
Four places the desk gets faster
Every one of them is a place a trader-operator currently loses money to paperwork, a stale assumption or a missed date.
See every cargo, once
Several hundred broker mails a day become structured orders and open positions. The same cargo from four brokers resolves to one opportunity with four sources, so the desk works the market instead of the inbox.
Price on what your ships actually do
Observed consumption for that vessel, observed port time at that berth, bunkers by grade and emissions costed per leg. Every input carries its source and its age, so a trader can see what the number rests on.
Open where the market clears
Weighted comparables from your own fixture history produce an opening band, an expected clearing level and the likely next counter. The comparables stay on screen, so the recommendation can be argued with.
Let agents do the chasing
Estimates built on arrival, offers drafted, chain terms compared, port accounts queried, bar dates escalated. The work arrives done. A named person decides what leaves the building.
The work arrives done
A circular lands and the estimate is already built. The trader reviews a priced, checked position instead of assembling one.
Inbox
Several hundred cargo and tonnage mails a day
Structured
Deduplicated across brokers, resolved to real entities
Matched
Physical fit and availability, with exclusions explained
Priced
Observed performance, real port time, emissions per leg
Checked
Chain terms compared before the recap is confirmed
Yours to send
A named person presses send. Always
Where the hours go today, and where they go once the desk stops re-keying
Why the cycle matters in a fast market
In a moving market the desk that answers first gets the business. Time from circular received to firm offer sent is the measure, and it is spent almost entirely on work nobody wants to do: reading, de-duplicating, re-keying, looking up what that berth usually takes.
- Capture runs continuously, so the position is current when a trader opens it.
- The estimate builds itself the moment a match appears.
- The offer is drafted from approved positions, in market conventions.
- The chain comparison runs before the recap is confirmed.
The same desk, with the system underneath it
The desk today
The same desk on Capesize
A trader reads the same cargo from four brokers and treats it as four opportunities.
One opportunity, four sources, confidence on every field.
The estimate runs on warranted speed and a default port time.
Observed consumption for that vessel, observed port time at that berth.
A term mismatch between the in and out charter surfaces at claim.
It surfaces before the recap is confirmed, with the exposure quantified.
The operator works the inbox in arrival order.
The worklist is ordered by money at risk.
Hours gained at sea disappear into anchorage time unnoticed.
They are predicted, accrued and attributed to the link that owns them.
A demurrage claim is served late, or served incomplete.
The pack is assembled from what the clause requires, inside the bar.
Six things the desk cannot watch, and the system never stops watching
Fifteen live voyages generate more signal than any operator can read. The system ranks the day by money at risk.
Hours found before they are lost
Favourable current moves a vessel faster than the commercial baseline. The gain is predicted before arrival, accrued against the link that owns it, and turned into a decision: take the earlier berth, or slow down and bank the fuel.
Mismatches caught at the only moment they are fixable
Ninety days to claim on the way in, sixty on the way out. The comparison runs before the recap is confirmed and raises a task the trader owns, with the exposure quantified.
Discrepancies found while the ship is alongside
The statement of facts is reconciled against independent position data in real time. A discrepancy found before the claim is presented is worth many times one found afterwards.
No claim lost to a time bar
Every bar date is derived from its own clause, not a default, and escalates as it approaches. Where a claim is held for another party's account, the escalation names them, because the loss would be theirs.
Port accounts challenged, not just paid
Every final disbursement account is compared line by line against proforma, tariff and your own history at that port. The query is drafted for the operator to send.
A result finance accepts
One voyage result, trued up after the claim settles, attributed to its drivers, with the corrected assumptions flowing back into the next estimate.
The operator's day, reordered
Email arrives in the order it was sent, which has nothing to do with what it costs. The worklist is ordered by value at risk, and every item carries the evidence that put it there.
Suppression is aggressive, and the dismissal rate is itself monitored. A list nobody trusts gets ignored, which removes the capability rather than the noise.
Fifteen live voyages, ranked by money at risk
- High
V-118Kamsarmax, 82k
Demurrage time bar in 9 days, pack missing two documents
- High
V-121Capesize, 180k
Third early arrival on this vessel, 33 hours absorbed at anchorage
- Medium
V-104Panamax, 76k
Final port account exceeds proforma by a material margin
- Medium
V-109Supramax, 58k
Statement of facts disagrees with position data on commencement
- Watch
V-112Kamsarmax, 82k
Hold cleanliness inspection not yet recorded, notice tendered
Suppression is aggressive, and the dismissal rate is itself monitored. A list nobody trusts gets ignored, which removes the capability rather than the noise.
A trader-operator makes money in three places
The three are routinely confused with each other. Separating them turns an argument between two desks into a calculation both can see.
Trading margin
The spread
What the tonnage cost to take in against what the employment earned. Won or lost at the moment of fixing.
Trade deskOperational alpha
The execution
Speed, bunker buying, port time, routing. Won or lost over the following six weeks, by the operator.
Operator deskRecovery
The claims book
Demurrage earned, demurrage paid, performance and off hire. Won or lost on documents and deadlines.
Operator deskWhere the system moves the needle
Claims served in time
Estimates on observed data
Chain mismatches caught at fixing
Gained hours attributed
Illustrative targets, not measured results. Each one is baselined from your own record before the build starts, and measured against that baseline afterwards.
Questions a desk head asks first
What does CapesizeOS actually replace?
The commercial system of record for a dry bulk trader-operator: cargo and tonnage capture, estimates, negotiation, fixture, the charter chain, voyage execution, port calls, laytime, claims and the voyage result. Technical ship management stays outside the perimeter by design, which is what keeps the product small enough to finish.
How is this different from a chartering system with a chatbot bolted on?
The agents work inside the same record the desk works in, so an estimate, a claim or a chain comparison is produced from your own data rather than described back to you. Each capability declares what it may do, the gate that converts a proposal into an action, and how it is undone.
Will an agent send an offer on our behalf?
No. Four limits are not configurable: no agent forms a contract, none signs or releases against a bill of lading, none moves money above a set value, and none dispositions a sanctions hit. The agent prepares the work completely and a named person performs the act.
What if our history is not good enough for the prediction layer?
That question is answered in week two, against the extracted record rather than the intention of it. The quote guidance layer sits behind an explicit readiness gate, and nothing in the first release depends on it. A disappointing answer reshapes the second release rather than derailing the programme.
How long before the desk sees anything?
Six weeks to one cargo running end to end alongside your current system, on real voyages rather than in a demonstration environment. Sixteen weeks to the old system going read-only. Cutover is a measured threshold rather than a date in a plan.
What happens to the numbers we already trust?
The calculations are rebuilt from the contracts and tested against voyages your current system already closed. Where the two disagree, the disagreement is investigated rather than assumed to be a new bug. A replacement that cannot tie to your existing numbers does not get adopted.
One cargo, end to end, in six weeks
Running on real voyages alongside your current system, with the claim served inside the bar and the voyage result agreed by finance. Then twenty voyages. Then the book.