Move goods, documents and moneyacross the border.
Purser is the system of record for the shipment file: one record from purchase order to bank realisation, every document projected from it.
The exporter authors the documents. The importer carries the penalties.
The gap between authorship and liability is the product. The exporter's anxiety is money that has not arrived; the importer's anxiety is a physical object burning money right now, in a yard, under someone else's control.
Money that has not arrived.
One entry, the whole docset. The shipment thread carries the canonical field set once; every invoice, packing list and declaration is a rendering of it, and every clock counts back from its own dates.
See Purser Outbound →The countdown · For importersA box burning money on a clock you did not start.
The bill of entry, the free time at the terminal, and what one more day costs. The proforma goes in, the checked declaration data set comes out, and your own broker files it.
See Purser Inbound →Two apps, one core, one agent.
The paperwork is the risk.
Five decisions carry the whole system.
Everything else in Purser is a consequence of these. Each one is load-bearing: remove it and the record stops being trustworthy.
One addressable object per shipment, like a pull request. Change the invoice number and every downstream artefact goes stale, with every recipient notified. Documents are renderings of the record, never independent files.
Versioned, effective-dated rulepacks. Every rule carries a citation; a rate change ships as a signed data publish, not a deploy; and evaluation binds to the shipment's own date anchor, never today.
No government submission ever, and no outbound message without a recorded human approval event. The agent drafts, checks and stages; a person presses send.
A per-shipment email address, a WhatsApp lane, and scoped, time-boxed, revocable magic-link grants. External seats are permanently free and unlimited.
A country-agnostic core; country knowledge enters as versioned modules. India is the first module, built deeply. The core itself names no country.
One entry.
The whole docset.
The exporter flow, end to end. The importer flow, the countdown, lives on the Inbound page.
Purser never files with a government portal. Your CHA or broker presses file.
Forward the purchase order, or enter it once
The shipment record opens with the canonical field set: parties, goods, values, dates. Everything downstream is projected from here.
The docset projects from the record
Invoice, packing list, declaration, each a rendering of the one record. Change a field and every stale artefact says so, to every recipient.
The rulepack fires, every verdict cited
Rules in force on the shipment's own dates, each verdict carrying its instrument and effective date. No citation, no claim.
Purser drafts, you approve
Checklists, chasers and gap explanations land in a review queue. Nothing leaves without a recorded human approval event.
The filing stages at awaiting approval; a human transmits
The checked data set goes to your CHA or broker on their free console. They press file, exactly as they do today.
Purser keeps the papers.
On every ship that ever sailed, the purser was the officer who kept the papers and the money. Purser's agent does the same for the shipment record: it reads, checks, drafts and watches the clocks, and hands every action to a human.
Ask what is blocking you. Get a cited answer.
Purser answers from the record, not from memory. Every line it produces carries the instrument and the date behind it, so an answer is something you can act on, and defend later.
- Composed checklists. What this shipment needs, with a citation on every line.
- Docset validation. Every gap explained with the instrument and date behind it.
- HS classification candidates. Proposed with reasoning, for a human to confirm.
- Provisional duty, RoDTEP and drawback. Computed from the rates in force on the shipment's own dates.
- Every statutory clock watched. Days remaining on each, counted from the shipment's own anchors.
- Drafts into a review queue. A human approves before anything is sent.
Rules are data.
Compliance knowledge lives in versioned, effective-dated rulepacks that compose in a fixed order. Every verdict names the instrument and the date it stands on.
observed shipping bill not filed · expected filed before let-export order
16 rulepacks loaded, about 180 cited rules across export and import, 6 industry and import packs at launch.
Free forever for counterparties.
The CHA, the forwarder, the bank and the buyer, each on a scoped, time-boxed, revocable lane. Per-shipment email, WhatsApp, magic links. The CHA who works for eleven exporters sees eleven scoped lanes.
Bridge them. Never make them log in.
Nobody outside your firm creates an account to work a shipment. Each counterparty gets a lane scoped to exactly the shipments shared with them, time-boxed, and revocable the moment the engagement ends.
- Per-shipment email. Every shipment has its own address; anything forwarded to it is filed on that shipment.
- WhatsApp lane. The channel your counterparties already answer on, attached to the record.
- Magic-link grants. Scoped, time-boxed, revocable. No passwords, no seats, no bill.
Trust, in writing.
Every event on the shipment record carries its actor, its channel, and a sha256 chain to the event before it. The trail is tamper-evident by construction, not by promise.
Every rendered document is hashed at render time, and the viewer serves the copy in circulation, not a fresh projection. What your counterparty saw is what the record holds.
No freight margin. No lending spread. Never sell to the intermediary. Never meter the document. Never operate a network we also compete on. Neutrality is policy, not posture.
A full export of the shipment record in an open schema, on demand. The exit is in the contract, so the record is yours in practice, not only in principle.
Row-level filters on every store, including vector memory. One tenant's shipments, documents and agent memory are never reachable from another's queries.
Every rule verdict and every Purser answer names the instrument and the date it stands on. A claim that cannot cite its source does not ship.
A document is never metered. An external seat is never charged.
One thread, three products.
The realisation rail
Export realisation, closed and certified, on a live DGFT integration.
Visit eBRC →ScripXScrip liquidity
Liquidity for RoDTEP and RoSCTL duty-credit scrips.
Visit ScripX →PurserThe thread that feeds both
One record per shipment, from purchase order to filing. The shipment object the other two rails read from.
You are here →Asked, and answered plainly.
No. Purser never transmits to a government portal. Every filing is prepared, checked against the rules in force on the shipment's own dates, and staged at awaiting approval. A human transmits it: your CHA or customs broker presses file, exactly as they do today, with a checked data set in hand instead of a folder of attachments.
No. Purser sells to the exporter and the importer, never to the intermediary. Your CHA keeps the filing, the relationship and the fee. What changes is what they receive: a checked declaration data set on a free, scoped console instead of a WhatsApp thread of PDFs. The CHA gets a console, never a bill.
No. Every counterparty works on a scoped, time-boxed, revocable lane: a per-shipment email address, a WhatsApp lane, or a magic-link grant. They see only the shipments shared with them, they never create a password, and external seats are free forever, without limit.
A new rulepack version is published as signed data, never as a code deploy. Evaluation always binds to your shipment's own date anchor, so a shipment is judged by the rules in force on its own dates and past periods never silently reprice under a new rate.
Yes, and in writing: a full export of the shipment record, every document, every event and every decision, in an open schema, on demand. The exit is part of the contract, not a support ticket.
India, deeply, today. But the core names no country: country knowledge enters Purser as versioned rulepack modules, so India is a country module, not the product. New corridors are new modules on the same shipment object.
No, by policy. Purser takes no freight margin, no lending spread, never sells to the intermediary, never meters a document, and never operates a network it also competes on. Neutrality is the point: the record is only trustworthy if the party keeping it has no position in the trade.
Own the shipment record.
Give away the document work forever to own the shipment record at the earliest possible moment.