“Vessel departed Antwerp 01.11.”
Freight forwarding operations
Turn shipping updates into actionable timelines.
BerthVellum helps freight forwarders turn scattered shipment emails and documents into one operational timeline. AI extracts milestones, reconciles conflicting updates and flags shipments that need attention — for operations staff to verify.
- Carrier · 2 Nov“MV Orion Bay sailed Shanghai 02/11.”
- Agent · 10 Nov“Box loaded on feeder at Tema 09/11.”
- Arrival notice (PDF)“ETA Apapa: 15-Nov-2026”
- Agent · 12 Nov“Feeder delayed, now expecting the 18th.”
- Departed origin2 Nov
- Transshipped · Tema9 Nov
- ETA Lagos15 or 18 Nov?
- Berthed—
- Customs released—
The problem
A shipment’s real status is spread across five inboxes and a phone call.
Carriers, agents, terminals and brokers each send their own update, in their own format, on their own schedule. Teams rebuild the picture by hand — and a missed handoff becomes storage charges, demurrage or an unhappy client.
“Northern Star don berth this morning.”
ETA Apapa: 15-Nov-2026
Release status: PENDING
“Customs release obtained for HLXU3304182.”
Milestone extraction
One timeline per shipment, with the source of every date.
BerthVellum reads correspondence and documents, extracts the milestones — departed, transshipped, ETA, berthed, discharged, released, delivered — and places them on one timeline. Every date links back to the email or document it came from, and operations staff verify it.
- Verified
- Sources disagree
- Not yet reported
Conflict detection
When two sources disagree, a person decides which is right.
ETA APAPA: 15-NOV-2026
extracted · 94% confidence“Feeder delayed by congestion, now expecting the 18th.”
extracted · 86% confidenceBerthVellum flags the conflict, shows both sources side by side and raises the shipment in the exception queue. The coordinator confirms with the agent or line, picks the correct update and records how they confirmed it. Nothing is overwritten silently.
Exception queue
Today’s follow-ups, most urgent first — with the reasons.
- BV-2045Adewale Foods · Ningbo → Lomé → Lagos (Tin Can)high
- Free time ends in 1 day+35
- Bill of lading not received — cargo is already at port+25
- BV-2043Okafor Building Supplies · Antwerp → Lagos (Apapa)medium
- Sources disagree on “Customs released”+30
- Packing list not received+10
- 1 low-confidence update not yet verified+5
- BV-2041Adewale Foods · Shanghai → Tema → Lagos (Apapa)medium
- Sources disagree on “ETA destination”+30
- Form M not received+10
- BV-2042Lumen Pharma · Mumbai → Lagos (Tin Can)medium
- Free time ends in 2 days+35
- BV-2046Kano Textiles · Chittagong → Colombo → Lagos (Apapa)medium
- Transshipment not confirmed+20
- 1 low-confidence update not yet verified+5
Computed from the fictional shipments in the desk, as of 14 Nov. Scores come from rules teams recognise; a learning model for delay prediction is added only as enough history builds up.
How it works
From inbox to timeline in five steps.
- 1
Forward
Shipment emails and documents are forwarded or uploaded for a set of workflows.
- 2
Extract
Milestones, dates and references are read from emails and documents.
- 3
Reconcile
Conflicting updates are flagged with both sources side by side.
- 4
Verify
Operations staff confirm, correct or reject each extracted event.
- 5
Act
Exceptions are queued by urgency; staff follow up and close them.
Ground rules
Built to support the desk, not replace it.
Staff verify every event
Extracted dates are suggestions until someone in operations confirms them.
No silent overwrites
When sources disagree, both stay on record and a person chooses.
No automatic messages
BerthVellum never emails clients, carriers or agents on your behalf.
Evidence travels with the date
Every milestone links to the email or document it came from.
Why it matters
Most trade moves by sea, and schedules slip.
Published figures from the sources linked. They describe the industry, not BerthVellum results.
Who it’s for
For the people who answer “where is my container?”
Independent freight forwarders
Many shipments, small teams, every client asking “where is it?”
Shipping agents
Updates from lines, terminals and principals to reconcile.
Import operations teams
Releases, documents and free time to watch across suppliers.
Organisation types are prospective customers; no affiliation is implied.
Pricing
Priced by active shipments.
Illustrative pricing for a proposed product, agreed with pilot customers.
Desk
$290 per month
- Up to 150 active shipments
- Milestone extraction
- Conflict detection
- 5 users
Operations
$790 per month
- Up to 600 active shipments
- Exception queue & alerts
- Document processing
- Unlimited users
Network
Custom
- Multiple branches
- Carrier & portal integrations
- Delay prediction as history builds
- Dedicated support
Proposed platform
Proposed architecture on AWS.
Not deployed. This website and the shipment desk run in your browser on fictional data.
- Amazon BedrockExtracting shipment updates
- Amazon TextractReading shipping documents
- Amazon SageMaker AIDelay prediction, as history builds
- Amazon S3Correspondence and documents
- Amazon DynamoDBShipment timelines
- AWS LambdaProcessing and alerts
The team
The founders of BerthVellum.
Founder & CEO
Chukwudi Okeke
Chukwudi has run forwarding desks where a single missed email meant demurrage, and started BerthVellum so every shipment has one timeline the whole team can trust. He leads product and forwarder partnerships.
Co-founder
Yetunde Balogun
Yetunde leads the platform: extracting milestones from correspondence and documents, reconciling conflicting updates, and the exception queue operations teams work from.
FAQ
Frequently asked questions.
Which shipping workflows does the first version cover?
A limited set of import workflows: departure, transshipment, arrival, berthing, discharge, customs release and delivery, from customer-uploaded correspondence and documents.
Does BerthVellum contact carriers or clients?
No. It organises and flags. Your team decides who to contact and what to say.
How does delay prediction work?
At first, from rules teams already use — conflicting updates, free time running out, missing documents. A learning model is added only once enough of your own history exists to test it.
Is BerthVellum live?
Not yet. It is a proposed product; the shipment desk is a prototype running in your browser on fictional data.
Pilot
Bring a month of shipment emails.
A pilot starts with correspondence and documents for a limited set of import workflows. Your team verifies every extracted event and compares the timelines and exception queue with what actually happened.
Or write to chukwudi@berthvellum.site — Chukwudi Okeke replies personally.