Aircraft parts traceability on a private blockchain

An aircraft parts traceability platform built on a private blockchain

AirpartChain is an aircraft parts traceability platform that turns the paper trail behind a component into a record nobody can quietly rewrite. It does two jobs that used to be separate. First it digitizes aircraft maintenance records with OCR and an AI layer, so a logbook page becomes structured, searchable data. Then it anchors those records to a private blockchain, so the history of a part stays tamper-evident once it is in the system. A document goes in, it is read and validated, and a fingerprint of it is committed to a ledger that every authorized party can check but no single party can edit.

The platform serves the part of aviation where a record isn't paperwork but proof. A repair that can't be documented might as well not have happened, because the aircraft can't show it's airworthy. A used part with a history nobody trusts is hard to sell and risky to fit. AirpartChain set out to fix the record layer underneath all of that, and brought Idealogic in to build it end to end: the web application, the UI and UX, the blockchain layer, and the AI and OCR pipeline that feeds it.

What makes the platform different from a document store is the assurance attached to every record. The history of a component travels with it, the proof of integrity comes from math rather than from trusting one company's database, and the records can move between an MRO, an operator, and a buyer without any of them holding an editable copy. That is the shift AirpartChain is built around: turning a part's history from a folder of scans into evidence that holds up when someone asks where the part has been.

Why aircraft parts traceability breaks down on paper

Aircraft maintenance runs on paper, and paper gets lost. Work performed on an airplane goes unrecorded when the documentation never makes it into a system, and an undocumented repair is a gap in the airworthiness record. Around that core problem sit two more. Spare-parts histories are scattered across sources nobody fully trusts, so the same used component can look airworthy to one party and questionable to another. And maintenance schedules built from manual data entry carry a quiet risk, because a typo in a record is not a clerical issue but a safety one.

The records themselves are the hard part. Maintenance history arrives as scans, faxes, and decades-old paper, and turning it into structured, verifiable data is slow manual work. The documents that carry the proof are specific and standardized: release certificates such as an FAA Form 8130-3 authorized release tag, logbook pages, and work orders. The FAA even publishes guidance, AC 43-9C, on what a maintenance record has to contain. So the information exists. It's just trapped on paper and impossible to verify at a glance.

Trust between parties is the second failure. When a part changes hands, the buyer inherits whatever records the seller kept, in whatever state they kept them. There's no shared source of truth, so each side reconstructs the history from its own copies, and a record that's been altered, on purpose or by accident, is hard to catch. The cost of a parts history that can't be defended is measured in grounded aircraft and disputed sales, not in inconvenience. The platform had to produce records that any authorized party could rely on without taking the holder's word for it.

What we built for aircraft records management

We built one platform that handles records management and traceability end to end, from the moment a paper document is scanned to the moment a buyer or an auditor verifies it. A maintenance document is read by OCR, validated by an AI layer, and turned into a structured record. That record is anchored to a private ledger, and from then on its history is provable. Live access and a machine-to-machine interface let existing systems read from the platform instead of waiting on paperwork.

These four pieces are one system, not four features bolted together. Digitization without an anchor would produce searchable records that are still editable. An anchor without digitization would have nothing to commit. The API matters because traceability only pays off when a maintenance system or a parts platform can ask for a part's history directly, which is why we built the platform to be read from, not just logged into.

OCR and AI: turning paper maintenance records into structured data

The ingest layer is where most of the difficulty lives, and it's the part comparable systems tend to skip. Getting from a scanned logbook page to a trustworthy record is two steps, not one. OCR pulls the text off the page, including handwriting and the cramped layouts of release certificates and work orders. That alone isn't enough, because raw OCR output is noisy and a wrong character in a part number or a date is exactly the kind of error a safety record can't carry.

So an AI layer built on OpenAI sits between extraction and storage. It validates and structures what OCR produced, checking the fields make sense before they become a record rather than trusting the scan blindly. A release tag has expected fields; a work order has its own shape. The model reads the extraction against what the document is supposed to contain and flags what doesn't fit. The output is structured aircraft maintenance records that can be searched, queried, and linked to a specific component, which is the difference between a folder of images and aircraft records management you can actually use.

This is also where the platform earns the right to anchor anything. There is no point committing a record to a tamper-evident ledger if the record was wrong when it went in. Cleaning and validating the data first is what makes the chain meaningful: the platform is anchoring a record it has reason to trust, not a guess scraped off a fax. The OCR-to-validation-to-anchor pipeline is the spine of the whole system.

On-chain and off-chain: how the chain of custody stays tamper-evident

The architecture rests on a deliberate split between what goes on the chain and what stays off it, and that distinction is the heart of the platform. The documents themselves are stored off-chain, where they can be indexed, access-controlled, and held like normal files. Only a cryptographic hash of each record, plus the custody events that link a part to its history, go on-chain. A hash is a fixed-length fingerprint of the exact content: change a single character in the underlying record and the hash changes completely.

There is a concrete reason to build it this way rather than putting everything on the ledger. Writing full scans to a blockchain would be slow and costly, and it would be a privacy problem, since a chain is hard to redact and aircraft records are commercially sensitive. Anchoring only the hash keeps the sensitive content private and the storage manageable, while still committing an immutable proof of the exact record that was filed. The chain carries the evidence; the off-chain store carries the document.

Verification is what turns that design into chain of custody software rather than a claim. To confirm a record, the platform re-hashes the stored document and compares it against the on-chain entry. A match proves the record hasn't changed since it was anchored. Nobody has to trust AirpartChain's own database, because the math does the trusting: anyone with access can check a part's provenance for themselves. That's the practical meaning of tamper-evidence here, and it's what gives blockchain traceability its weight in a dispute, where the conversation moves from argument to a verification anyone can run.

Architecture: a private Hyperledger ledger with an API gateway

The ledger is a private Hyperledger Besu chain with access control, and the choice of a permissioned network over a public one is deliberate. The parties to an aircraft record are known to each other, and the records are sensitive, so a public chain would be the wrong tool: it would expose data and bring throughput and cost limits that operational records can't afford. A private chain keeps who can read and write under control while still giving every authorized party the same immutable view. This is aviation blockchain used for what it's actually good at, which is shared trust between named participants rather than open, anonymous access.

On top of the ledger sits a real-time data layer and an API gateway. The gateway is what lets the platform be useful inside systems that already run. A maintenance system or a parts platform can request a component's history through a machine-to-machine interface instead of routing a person to a separate app, and live data means the answer reflects the current record rather than yesterday's export. Records management is only worth the effort when the records are reachable from where the work happens, so the platform was built to be queried, not just to store.

Underneath all of it, the security posture and the records model were designed together rather than retrofitted. Access control, the off-chain store, the hashing pipeline, and the ledger are part of one design, which is why the privacy model and the traceability model line up instead of fighting each other. The result is a stack where each piece, the OCR and AI ingest, the off-chain document store, the private Besu ledger, and the API gateway, earns its place by removing a separate tool a maintenance or parts business would otherwise have to run and reconcile on its own.

Results: provable aircraft parts traceability for MROs, operators, and buyers

AirpartChain is live in production. The shift it delivers is in what a maintenance record now is: not a scan in someone's email, but a structured entry with a tamper-evident history that any authorized party can verify. Maintenance planning works from data instead of recollection, and the spare-parts side of the market gets documentation it can actually trust. Aircraft parts traceability stops being a paper chase and becomes a lookup with proof attached.

The same platform serves several kinds of participant, because the record underneath is the same no matter who needs it. The grid below shows what changes for each, while the digitization, the private ledger, and the verification stay constant beneath them. What differs is the question each party brings to the history, not the machinery that answers it.

What started as a way to get maintenance work recorded has become the proof layer the market trades and audits on. AirpartChain sits in our aviation and aerospace practice next to ePlaneAI, the parts marketplace. One platform trades the components; the other proves where they have been. The provenance machinery comes from our blockchain development practice, and the records and AI work follows the same discipline as the rest of our custom software development and AI integration: model the records and the regulation first, then build the system around them. If the vocabulary in this corner of aviation is new, our primers on what MRO means in aviation and the aerospace supply chain cover the ground around it.