Freightville

Shipment verification network · Pilot program

Both ends of every delivery, on one ledger.

A sensor on the truck records what left. A sensor on the tank records what arrived. Freightville writes both readings to a tamper-evident ledger, compares them within your tolerance, and alerts the moment they don't agree.

Verification record FV-SHIP-2231

Dispatched · AquaHaul

9,980 L

truck meter AH-FLOW-114-1 · sensor-attested

Received · Ridgeline Farms

9,942 L

tank meter RF-TANK-07-A · sensor-attested

Δ 0.38% · tolerance 0.75% (lane) block #1 · 3/3 validators
Verified0.38% ≤ 0.75% · on ledger

How it works

Measure at both ends. Agree in the middle.

No trust in either party's spreadsheet, and no arguing after the fact. Each reading is signed by the organization that took it, sealed into a block, and reconciled automatically.

SENSE

Meters read the transfer

The dispatching meter and the receiving meter each record the quantity — flow meters, tank level sensors, scales, or a manual reading from the field app when hardware is down.

SIGN

Each side signs its own reading

Readings are cryptographically signed with the recording company's key. A shipper can't write a receiver's numbers, and nobody can rewrite a reading later.

SEAL

Validators seal the block

Named validator organizations countersign each block. A block is final only with a two-thirds majority — one party alone, including Freightville, can't alter the record.

RECONCILE

Compared within your tolerance

Out and in are matched per shipment and checked against the tolerance you set — per contract, per lane, per commodity, or a company default. Breaches alert immediately.

ShipmentRouteOutInΔToleranceStatus
FV-2231AquaHaul → Ridgeline9,980 L9,942 L0.38%0.75% · laneVERIFIED
FV-2232AquaHaul → Ridgeline5,000 L4,968 L0.64%0.75% · laneVERIFIED MANUAL — REVIEW
FV-2233AquaHaul → Ridgeline12,000 L11,280 L6.00%0.75% · laneVARIANCE
FV-2230AquaHaul → Ridgeline8,000 L24 h windowUNPAIRED

Output of the reference implementation in this repository — run python3 chain/demo.py to reproduce it, including the alert log and a rejected tamper attempt.

The ledger

A blockchain that spends signatures, not electricity.

Freightville runs a permissioned chain with Proof-of-Authority consensus. Validators are named, contracted organizations — not anonymous miners — because in freight, identity is the point.

> ⅔

Quorum finality

A block is final only when more than two-thirds of validators have signed it. No single company — Freightville included — can alter or censor the record alone.

~0 kWh

No mining, ever

Consensus work is an Ed25519 signature. The chain's energy budget is a handful of ordinary servers — the same class of machine that runs your dashboard.

Anchored

Publicly tamper-evident

The chain's Merkle root is published to an external public record on a schedule. Even full validator collusion can't rewrite history older than the last anchor without detection.

Architecture

The chain is the record. The database answers to it.

A ledger is a terrible query engine, and personal data must never be immutable. So Freightville puts each kind of information where it belongs — and keeps a rule for when they disagree.

On the ledger

integrity is the product
  • Signed measurement events — every OUT and IN reading, with device ID, trust tier, and meter certification.
  • Alerts and their resolutions — not just that a variance fired, but what was done about it.
  • Tolerance changes — if a tolerance widens the day before a shorted delivery, the record shows it.
  • Role grants and staff access — every Freightville read of customer data is written here, visible to that customer.

In the database

operational · queryable · erasable
  • Accounts and sessions — users, passwords, API keys, support tickets.
  • Personal data — names and emails stay erasable, as privacy law requires. The ledger holds org and device IDs, never people.
  • Device registry and dashboard state — configuration that changes daily and needs no notary.
  • Analytics projections — variance by lane, by month, by commodity, answered in milliseconds, not by walking blocks.

Photos follow the same principle. A meter-face photograph from the field app lives in object storage; its SHA-256 fingerprint lives on the ledger. Anyone can prove the photo is the exact one recorded at delivery — without a multi-megabyte image bloating the chain, and while the file itself stays deletable.

The dashboard reads a projection built from the chain — fast to query, rebuildable from the ledger at any time, and continuously checked against it. If they ever disagree, the chain wins, and the disagreement is itself a security alert.

Standards & protocols

Speaks the standards your hardware already speaks.

Every source — any manufacturer, any transport — is converted at the edge into one canonical event format. New protocols are new adapters; nothing downstream changes.

Event format
GS1 EPCIS 2.0 (ISO/IEC 19987) — the international standard for supply-chain event data. Every reading answers what, when, where, and why.
Units
UN/ECE Recommendation 20 codes (LTR, KGM, MTQ…), normalized at the adapter so mismatched units are impossible past the edge.
Sensor transport
MQTT (ISO/IEC 20922), OPC UA (IEC 62541), LoRaWAN, and HTTPS/JSON — with an adapter SDK for manufacturer-specific protocols as they appear.
Meter certification
Each reading carries its instrument's legal-metrology certification (OIML R117, OIML R49, NTEP), so a disputed number traces to a certified — or uncertified — device.
Interoperability
Verified events export to ANSI X12 214 / EDIFACT status messages for the TMS and ERP systems you already run.

Accounts

The same roles at every scale.

Companies join as shippers, receivers, suppliers — or all three, since the same firm often receives at the depot and ships onward. Inside each company, three admin roles mirror the platform's own, scoped to that company alone.

RoleFreightville staff — all accountsYour company — your account only
Full Admin Everything, including validator keys, billing, and role grants. Sensitive actions require two Full Admins. Everything in your company: users, API keys, devices, role grants.
Site Admin Configure tolerances, adapters, sites, and devices across all accounts. Configure your sites, devices, tolerances, and integrations.
Support Admin Read access for support — no configuration writes, no key access. Every read of customer data is itself written to the ledger. Read access to your company's records and support tickets.

Platform-wide access is audited by design: customers can see exactly when Freightville staff viewed their data, because those reads are on the ledger too.

In the field

When the sensor is down, the delivery still counts.

The Freightville field app runs on any phone — it installs from the browser, works offline in yards with no signal, and syncs when coverage returns. When a meter is dead or a site has no sensor, the driver or the receiving operator records the reading by hand: quantity, a photograph of the meter face or gauge, GPS, and a timestamp, signed with their account key.

Manual entries are never laundered into looking like sensor data. Every reading carries its trust tier on the ledger, and reconciliation policy can treat tiers differently — a variance that clears on two attested sensors can still queue for review when a manual entry is involved.

T1

Sensor-attested

Signed inside a tamper-evident device with a registered key.

T2

Sensor-reported

Device reading relayed and signed by the operator's gateway.

T3

Manual, corroborated

Hand entry with meter photo, GPS, and timestamp.

T4

Manual

Hand entry alone — accepted, flagged, and reviewable.

Partners

Watched by people who watch networks for a living.

A verification network is only as good as its uptime — a reading that never arrives looks exactly like a delivery that never happened. We don't improvise that discipline in-house.

Ntersect Technology Partner

Ntersect provides Freightville's network observability. Their Ntrospect platform, built on OpenTelemetry, watches the paths our readings travel — sensor gateways, ingestion adapters, and validator nodes — with anomaly detection that surfaces degradation before it becomes a gap in the ledger.

When an UNPAIRED alert fires, the first question is whether the delivery failed or the network did. Ntersect's telemetry is how we answer it with evidence instead of a guess.

ntersect.com →

Pilot program

Put one lane on the ledger.

The pilot starts small on purpose: one commodity, one route, both ends instrumented, your tolerance. If the numbers don't earn their keep in ninety days, walk away with your data.