Using two technologies: digital product passports, and blockchains.
app.transparium.com · Figures dated August 2026 · S speaker notes · ? shortcuts
The problem
Precision manufacturing runs on documents — and the documents are the weakest part of the system.
Technology one
A Digital Product Passport carries a certificate of ownership and authenticity that any counterparty can validate — built to the EU DPP standards, so it is a recognised record rather than a private file.
Technology two
The specification is committed to a blockchain. The passport carries only the reference — so the record stays small, and the document stays provably unaltered.
The two together
Neither technology is sufficient alone. A passport with no anchored specification is a claim; an anchored specification with no passport is a file nobody can attribute.
Before the detail
What provenance failure actually costs, and why the technical record expires long before the part does.
The problem · provenance
Between January 2019 and July 2023, AOG Technics sold over 60,000 engine parts into the CFM56 supply chain with forged Authorised Release Certificates — one man in a home office in Surrey, doctoring genuine certificates on his own computer.
He was caught because an engineer at TAP Air Portugal saw a part that looked worn while its paperwork said new, and phoned Safran.
Say the whole truth: the FAA found no unsafe condition warranting an Airworthiness Directive (SAIB 2023-15). Nobody died. The director was jailed for 4 years 8 months in February 2026. The cost was £39.3m and an industry-wide records audit — a systemic and economic failure, not a crash.
The problem · where the gap actually is
CFM's audit found the fraud lived entirely in bolts, nuts, washers, dampers, seals and bushings — the batch-released commodity hardware nobody serialises.
This is why the worked example later in this deck is a fastener lot and not a turbine disc. The disc is already solved. The screw is not.
The problem · how trust propagates
The problem · longevity
43.10 consequence above is the rigorous version.
The problem · the obvious objection
It largely did, on paper. Chapter 16 defines a PKI-signed, tamper-evident electronic release certificate — accepted by the FAA, EASA and Transport Canada — where each eForm references the one before it, building a verifiable chain of custody.
Airbus, Boeing, GE, Pratt & Whitney and Rolls-Royce were on the project team. It has existed since the mid-2000s.
What the standards actually require, what is genuinely mandated for you today, and where a ledger earns its place.
The approach · regulatory reality
Indicative dates. Commission planning dates, not legal deadlines, and they have slipped before.
The approach · the standards
CEN/CENELEC JTC 24 delivered eight EN standards under mandate M/604. Six were cited in the Official Journal on 15 July 2026 and carry presumption of conformity — and they describe, almost exactly, the problem this deck opened with.
| Standard | Covers | Why it matters to a parts manufacturer |
|---|---|---|
| EN 18219 | Unique identifiers | Identity at model, batch or item level — batch is explicitly permitted |
| EN 18220 | Data carriers | QR, Data Matrix, NFC and RFID — Data Matrix named for direct part marking |
| EN 18221 | Storage, archiving, persistence | The multi-decade retention problem, as a standards requirement |
| EN 18239 | Access rights, security, confidentiality | Exists because commercially sensitive supply-chain data is the hard part |
| EN 18246 | Data authentication and integrity | Whether a passport can be trusted not to have been altered |
The approach · architecture
The approach · the honest boundary
A ledger guarantees a record was not altered after it was written. It guarantees nothing about whether the record was true when written. AOG Technics did not tamper with anyone's records — it invented its own, from scratch, and they were internally consistent.
A tamper-proof forgery is still a forgery. Any pitch that stops at "immutable" has not addressed the failure it claims to fix.
Four steps, and what each one actually writes. Nothing here is a mock-up — this is the shipped product.
How to use it · overview
The same four steps are what a university uses for degrees. What makes it an aviation DPP is the content of the certificate and the identity of the issuer — not a different product.
How to use it · step 02
The contract is the container. Everything issued under it inherits its issuer identity, its access rules and its verification status — so the boundary you draw here is the boundary a customer sees.
How to use it · step 03
A certificate is a name and a value. The name is how you find it; the value is what you are attesting. Both go on-chain.
HST12-8-6/000417sha256:… — the hash on-chain, the document in the content store.
// what you attest, on-chain
part_number "HST12-8-6"
serial "000417"
lot "24B"
material "Ti-6Al-4V"
spec_sheet "sha256:9f2c4a…e71b"
release_ref "EASA-F1-2026-04471"
manufactured "2026-03-11"
Deliberately not on-chain: anything personal, anything a customer could demand be erased, and the Form 1 PDF itself. Those live in the content store, committed to by hash. That split is what keeps the design compatible with a right-to-erasure request.
How to use it · the carrier
Every certificate produces a URL and an SVG QR code. The SVG matters: it scales to a 3 mm laser mark without resampling.
// the short form — what goes under a Data Matrix or QR on the part
https://app.transparium.com/reader?address=0x7Ac4…19bE&item=HST12-8-6/000417
// the long form — carries the read ABI, so it resolves without our lookup
https://app.transparium.com/reader?address=0x7Ac4…19bE&abi=%5B%7B%22name%22…¶ms=HST12-8-6/000417
How to use it · step 04
A tamper-proof record issued by an anonymous wallet proves nothing. The identity layer is what turns it into evidence.
How to use it · control
"Active verification" means the record has an owner, and the owner is enforced by the contract rather than by a login screen.
One specification, a thousand parts, and the arithmetic shown rather than asserted.
Worked example · the unit of provenance
You could give every screw its own passport. You should not, and the reason is not cost.
QTY=4 against a single lot number and no serials at all. Fastener traceability runs to the material heat, not to the item.
Note for standard parts: a fastener designated a standard part under EASA 21.A.307(b)(1) needs no Form 1 at all — a dated delivery note naming the part and traceable to the manufacturer suffices. A passport is an addition to that regime, not a substitute for it.
Worked example · the shape
Worked example · the arithmetic
Two things get conflated in every blockchain pitch. Separated here: what the network charges, and what we charge.
| Operation | What goes on-chain | Network cost | Charged as |
|---|---|---|---|
| Deploy the programme contract | Contract bytecode — once per programme | cents | One-off setup |
| Anchor the specification | A 32-byte SHA-256 hash, in contract storage | ~$0.05 | Per document |
| Issue the lot passport | Lot identity, material heat, release reference | ~$0.05 | Per lot |
| Optional: 1,000 item records | One commitment covering the whole batch | ~$0.05 total | No incremental charge |
| Network cost for the entire lot | Four writes, permanent, publicly verifiable | under $0.25 | — |
Basis: ~47,300 gas/anchor; mainnet base fee 0.118 gwei at block 25,789,431, 19 Aug 2026; ETH ≈ $1,935. Re-date before every presentation.
Worked example · the payoff
The four objections worth raising before someone in the room raises them.
Limits · 01
If the document body is deleted, the on-chain hash still proves that a file with that exact content existed at that time — and nothing more. It will not give you the file back.
Limits · 02
Arweave will store a whole 10 MB document permanently for around twenty cents. If your question is purely "what is the cheapest permanent byte", we are not the answer, and pretending otherwise wastes your time.
Limits · 03
Three things on Ethereum are called permanent and only one of them is. This is the distinction that decides whether a record is still readable in 2050.
| Where the data sits | Retention | Actually permanent? | We use it for |
|---|---|---|---|
| Contract storage variable | In the state trie every node must carry to validate the next block | Yes | Issuer identity, and the records that must last |
| Event log | In transaction receipts — within the scope of Ethereum's history-expiry work | Committed, but retrieval not guaranteed | Cheaper high-volume writes |
| Blob (EIP-4844) | 4,096 epochs = 18.2 days | No — pruned | Nothing. Deliberately. |
Why this is on a slide at all: blobs are the cheapest way to put bytes on Ethereum, and every cost-driven design lands on them. Any vendor quoting you a very low per-megabyte on-chain price should be asked how long the data survives. If the answer is a blob, it is eighteen days.
Limits · 04
Blockchain-anchored product passports are established practice in other sectors. Claiming to have invented the category would be checkable in about thirty seconds.
Limits · 05
Aerospace part data is export-controlled, design-approval-holder proprietary, and contractually restricted. "Put it on a public chain" is, correctly, the first thing your legal team will refuse.
Where this goes
Commodity hardware — the parts nobody serialises — is where the documentation fraud actually happened.
A verified issuer, a content hash, and a public reader that answers in a scan rather than an email thread.
One part programme, eight weeks, and one customer willing to scan a label at receiving.
app.transparium.com · All figures dated August 2026 and sourced in the speaker notes · S speaker view · ? shortcuts