A physical product can become a persistent digital object. A QR code, an NFC tag,
a barcode, a serial number: each is one doorway into the same identity. EOS is not a code
platform; it is the asset identity layer above your systems, and once the identity exists,
EOS can reconcile everything the enterprise knows about that object across its whole life.
The demonstration below runs on a synthetic tenant, in this page, on your device. Systems of
record stay authoritative throughout; EOS reads, reconciles, and routes, and never becomes
one.
Make the doorway
Enter an asset identifier and the code is generated here, in this page, on
this device. Nothing is sent anywhere. The estate deploys air-gapped and makes
no external calls at runtime, and a generator that called a third-party API
would break both.
What the code carries. A pointer to an asset identity, not
the asset's data. EOS reads from the systems that own the asset and reports
where they disagree. A scan resolves to that view, with the evidence chain and
the named person who can settle a contested figure. Scanning creates nothing,
changes nothing and authorises nothing.
Higher correction survives more damage and produces a denser code. A tag
going onto equipment in a yard wants H. A code in a document wants M.
PHYSICAL OBJECT → QR / NFC / RFID / SERIAL → PERSISTENT ASSET ID → LIFECYCLE GRAPH + RECONCILIATION + PROVENANCE + AUTHORITY → HUMAN + AGENT + WORKFLOW → DECISION, ROUTED TO THE OWNING SYSTEM
“Wait, if every product had an identity…” then a scan stops being a lookup and becomes an address.
“What if the identity persisted through the lifecycle?” Then history becomes structure: the lifecycle graph.
“What if the manufacturer, supplier, retailer and customer all contributed information?” Then no one owns the truth alone, and no one overwrites anyone.
“What if you could reconcile that information?” Then disagreement becomes visible, authority becomes nameable, and decisions carry evidence.
That last question is not hypothetical. Reconciliation is what EOS is. Four rungs up the ladder, you arrive at the bigger question:
What happens when an ordinary physical product becomes digitally identifiable throughout its lifecycle?
It stops being a row in somebody’s system and becomes a node that everything true about it
can attach to. One persistent identity, and thirteen dimensions of enterprise reality hanging
off it as sourced, reconcilable edges. That structure has a name: an asset graph. The rest of
this page is that graph, running.
Provenance. Every assertion carries source, document, clause, and verification date. Live on every claim above and below, the warranty chip, the attestation card.
Ownership. Title, custody, and transfer history join the same identity as sourced events. Attaches the same way as every edge here; custody chains already run on each asset.
Contracts. The clauses that govern the asset, warranty terms, compliance duties, price. Live in both warranty disagreements, clause 7.2 and clause 5.1, authority named.
Transactions. POs, requisitions, orders, invoices, credits, each tied to the identity they touch. Live in the planning panel, PO 77120, requisition 88231, the 1,200-unit order.
Warranty. Coverage, its clock, and the boundary where systems disagree about both. Live twice, the drive’s expiry conflict and the caliper’s clock-start conflict.
Service. Work orders, findings, measurements, the actor and the document behind each. Live in the caliper’s service records, wear 40 to 65 percent across two visits.
Location. Where the records say it is, and where the last physical scan says it is. Live in the caliper’s position mismatch and the drive’s rack agreement.
Condition. Test results, wear, telemetry states, condition as evidence, not opinion. Live in the end-of-line PASS log and the measured pad wear.
Financial information. Acquisition through lifecycle cost, valuations, and the registers that disagree. Live in every lifecycle value table and the robot arm’s finance agreement.
Compliance. Certificates, declarations, attestations, each with expiry and requiring clause. Live in the supplier-compliance signal and the committed attestation.
Environmental information. Material content, recycled share, origins, recovery value, the circular chain. Live in the 64 percent recycled alloy, origins, and the recovery credit.
Usage. Runtime, duty, and rate from the machines themselves, continuous truth. Live in the IoT line telemetry, 138 per day measured against 150 planned.
Replacement history. Lineage, not deletion: cores, remanufacture, and the position’s inheritance. Live in the caliper’s replacement lineage and RMA custody scan.
The graph is why every question below has the same answer. Traceability, recalls,
compliance, reverse logistics, planning, consumer verification: each is one traversal of the
same thirteen edges, and every edge carries its evidence.
Scan an asset
Two serialized assets from the synthetic tenant. Selecting one is the scan.
Every demonstration here follows the estate\u2019s six-beat standard: the enterprise problem, the discrepancy, what it costs, what the producer discovers, what changes, and the outcome. Every figure on this page is SYNTHETIC and labeled. When EOS runs this on your records, each assertion arrives with its source, document, clause, and verification date.
Passport
From part to application: the aftermarket chain
QR → PART IDENTITY → APPLICATION, THE VEHICLE OR EQUIPMENT → INSTALLATION → WARRANTY → SERVICE → REPLACEMENT → AFTERMARKET AND SECOND LIFE
Equipment identification. The part and the equipment it serves each hold a persistent identity, and the binding between them, which unit, which position, since when, is itself a recorded, sourced fact.
Installation history. Who installed it, when, under which work order, to which torque spec, with the document attached. The warranty clock and the next service both hang off this event, which is why it carries provenance.
Service records. Every service event joins the same graph with actor, findings, and measurements, so wear trends and the replacement forecast read from evidence rather than memory.
Asset tracking. Where the records say the part is, and where the last physical scan says it is, reconciled continuously; the third asset above shows what happens when they differ.
Replacement. Replacement is lineage, not deletion: the outgoing unit's identity persists through core return and remanufacture, and the incoming unit inherits the position's history.
Configuration. As-designed, as-built, as-maintained: three configurations that drift apart in every fleet. EOS holds them side by side, names the authority, and flags the drift, because torque specs, recall scope, and the next work order all key off what is actually bolted where.
Verifiable identity, and the layer underneath agentic commerce
Verifiable product identity. Each party attests to its own leg, the manufacturer to manufacture, the installer to installation, the servicer to service, as signed, independently checkable claims aligned with the W3C Verifiable Credentials standard. Identity that travels across companies without a phone call to verify it, and without anyone rewriting anyone else’s records.
Agentic commerce. An agent reordering the brake pad should not guess from one system. It asks EOS to establish state first, part, position, wear trend, warranty status, approved supplier, contract price, and receives the reconciled answer with its evidence. Then it prepares the order for the procurement authority to approve in the procurement system. Agents transact on established truth; EOS is the layer underneath.
Consumer verification: the QR as a cryptographic doorwayCOMMITMENT CHECK IS REAL
The same QR that opens the passport can hand the consumer a verifier. The
manufacturer commits its attestation, component sources, material content, countries of
origin, specification conformance, as a digest published to a public, append-only ledger
that neither the company nor any supplier can rewrite. Anyone can recompute the digest from
the disclosed claims and compare; and for what should stay private, zero-knowledge predicate
proofs let the consumer verify a statement about the committed data without seeing the data
itself.
Anchored digest, SYNTHETIC ledger entry for this demonstration: 868e8f2ed94cb8b7313f074eaaeb0ace9bd6d42df5cb4302a2244b98e3ec72fb
ZERO-KNOWLEDGE PREDICATES MODELED
Recycled content of the housing alloy is at least 60%VERIFIED
The consumer learns the statement is true. The exact percentage, the assay document, and the supplier stay private.
Every component’s country of origin is within the approved listVERIFIED
True or false, nothing more. Which countries, and the route, stay private.
End-of-line test conforms to specification TS-88VERIFIED
Conformance without the raw test log.
The commitment check on the left is real: your device just ran it. These
predicate rows are MODELED: at go-live they run as zero-knowledge proofs against the same
committed attestation, so the statement verifies while the data stays sealed. Nothing on
this card asks the consumer to trust the company’s word; the whole point is that they
do not have to.
Plug it in: days, not a program
1 · Connect, read-only. Credentials to the systems you already run, ERP, WMS, EAM, order and service systems, read-only from the first minute to the last. Nothing is replaced, nothing is migrated, and every system of record stays exactly as authoritative as it is today.
2 · Identities mint, codes generate. Every asset, lot, and serialized item in scope receives its persistent identity, and the QR sheets and labels generate from it, the same encoding the live generator below runs on your device.
3 · Passports go live. Scans resolve, the graph populates from your own records, disagreements surface with their authorities named, and the six-beat demonstrations above become your assets instead of synthetic ones.
What it takes: read-only credentials and a named authority table. What it does not take: a
migration, a rip-and-replace, or a services program. The estate’s standing rule holds
end to end: EOS reads, reconciles, and routes, and never becomes a system of record.
The commercial layer: every scan is a channel
The same QR that verifies also serves. Once the identity exists, the brand owns a channel
printed on the product itself, and the lifecycle graph makes it timely instead of generic.
Each capability below arms per tenant at pilot; the demonstrations above are the machinery
they run on.
Specs on demand. The scan opens the exact specification for this unit, this revision, this configuration, not the closest PDF on a support site.
Consumer replacements. Scan the worn part and the compatible replacement is already resolved, the razor-and-blade motion for any serviceable product, checkout on the same one processor every door runs.
Complementary offerings. What pairs with this unit, accessories, consumables, the next product in the system, offered from the graph rather than guessed from a segment.
Targeted promotions. Lifecycle-timed, not calendar-timed: the service-due nudge, the warranty-expiry upgrade window, the loyal-owner offer, each triggered by the asset’s own state.
Returns. The scan initiates the return with identity attached, RMA, custody, disposition, and credit reconciled against what physically arrives, the reverse-logistics rails above.
Warranties. Registration in one scan at install, claims with the evidence already assembled, and the clock disputes the demonstrations above resolve before they cost anyone a denied claim.
Recalls. Reach the actual owners, not a press release: the identities know where they are, the drill above scopes the lot, and the notification lands on the product in hand.
Marketing by lifecycle stage. Content that knows whether it is speaking to a prospect, a new owner, a mid-life operator, or a second-life buyer, because the graph does.
Generate a code, on your deviceTHE ENCODER IS REAL
Type a payload, an asset address, a serial, anything up to 78 plain characters, and your
device encodes it, a complete symbol with error correction, no server involved. The encoder
was verified by an independent decoder before it shipped: format bits checked, all twenty
error-correction syndromes at zero, and the payload read back out matching the input,
byte for byte. Tenant deployments mint these at scale, sheets and labels included.
Production planning, reconciled: component and finished-good truth
Two physical signal layers feed the planning stack: QR and barcode scans as event truth,
receipts, kits, cycle counts, genealogy binds, and IoT telemetry as continuous truth, line
rate, work in process, machine state. EOS holds them beside the systems that plan, promise,
and buy, and reports where they disagree. The fourth asset above runs the whole panel live.
Production scheduling. The MPS says what the line will do; the line’s own telemetry says what it has been doing. EOS keeps the schedule honest against measured rate, changeovers, and downtime, before the gap becomes a missed promise.
Customer orders. Every promise checked against producible truth, allocated capacity, reconciled inventory, and inbound supply, so available-to-promise means available in physics, not just in the planning table.
Requisitions, demand, and MOQs. The req that MRP raised, the demand that justified it, the minimum the supplier imposes, and the inventory the scans actually found, reconciled before the release, with the trapped cash of the MOQ bump priced in the open.
Line capacity. Planned rate versus measured rate, by line and shift, from the IoT log rather than the standard, so capacity claims carry evidence and the expedite decision reaches the right owner early.
Inventory levels. ERP on-hand versus the last physical scans at component and finished-good level, variances flagged with the records attached, the same discipline the homepage’s worked example runs at network scale.
Lead times. The master-data lead time versus the median the receiving scans actually recorded; when they drift apart, every MRP run plans the gap, and EOS names it with the receipts as evidence.
Maintenance, equipment, and tooling: the estate you already own
Maintenance management. Preventive schedules against actual service events, calendar-based and meter-based side by side, with the work-order feed as the configuration source the caliper case established.
External service. Third-party providers contribute service events under the same identity, signed and sourced, the four-party contribution model extended to the people who turn the wrenches.
Equipment management. The equipment registry as identities, fleet unit FLT-208 already is one, with utilization from telemetry rather than assumption, and every part binding recorded.
Tooling, internal. Dies, molds, fixtures, and jigs with calibration due dates, PM history, and shot or cycle counts from the machines, tracked like the capital they are.
Tooling, external. Company-owned tooling in supplier custody, the estate nobody audits: who has it, its attested condition, its consumed life, and its shot ledger reconciled against what actually shipped back. The fifth asset above runs the whole case.
Amortization. Tool cost recovered through piece price against a unit base, with the finish line watched: over-recovery surfaces the day the base completes, not the year after.
Depreciation. Book schedules against usage truth: straight-line versus units-of-production, idle assets still depreciating, fully depreciated assets still producing, and condition signals feeding impairment review instead of surprise.
Landed cost and cross-border truth
Every border crossing adds a system that can disagree with the one before it.
The sixth asset above runs the case live. All figures SYNTHETIC, illustrative rates and
codes, not a claim about any real tariff schedule.
Tariff and HS classification. The declared customs code reconciled against what the engineering BOM and the physical specification actually describe, so a supplier substitution cannot silently drift the classification, and the duty rate, out from under the product.
Duty and freight rollups. Standard cost reconciled against the true landed cost, invoice plus freight plus duty, so margin is never quietly overstated by a cost feed that stopped updating.
VAT, as a working-capital fact. Import VAT is usually recoverable, so it is not a cost, it is a timing problem, cash out now, cash back later. EOS carries it as its own line rather than burying it in landed cost, and ties directly to the cash simulator\u2019s terms-and-timing levers.
Trade-compliance evidence. The customs entry, the BOM revision, the classification rule, and the review date, all in one graph, so a classification decision is defensible on the day an auditor asks rather than reconstructed after.
Tariff exposure and opportunity. The same reconciliation that finds back-duty exposure also finds the reverse: a classification filed conservatively when the correct code is actually lower, or duty-drawback and free-trade-agreement eligibility sitting unclaimed because no one is watching for it.
The questions buyers ask, answered by the same architecture
“Could this be used for traceability?” Yes. The lifecycle graph read in both directions: backward from a unit to its lot, production run, supplier, and inputs; forward from a lot to every unit, location, and customer it reached. Most traceability systems assert a chain; EOS shows the source behind every link and reconciles the custody records that disagree.
“What about recalls?” Run one. A recall is a traceability query under time pressure: which units carry the lot, where each one is now, which customers hold them, which are already serviced or replaced, and the variance between records and reality. The drill on the first asset above answers with the evidence a regulator asks for, in seconds instead of a war room.
“What about product passports?” Already law. The EU’s battery passport requirement and the broader Ecodesign product-passport program are phasing in category by category, and a passport whose every claim carries source, document, clause, and verification date is what those regimes are actually asking for. Conformance mapping arrives with the first regulated pilot, and is never claimed before it.
“What about supplier compliance?” Certificates, declarations, and audit evidence live in the same graph as the parts they cover: mill certificates, origin declarations, restricted-substance and conflict-mineral attestations, each with its source, its expiry, and the contract clause that requires it. EOS reconciles what the contract demands against what the supplier filed and what the physical evidence supports, flags expiries before they lapse, and, through the verification layer above, lets a supplier prove a compliance predicate without opening its commercial books.
“What about reverse logistics?” The return path runs under the same identity as the outbound one: RMA authorization, custody scans back through the network, the disposition decision, restock, remanufacture, or scrap, routed to its owner, and the credit reconciled against what physically arrived. The caliper above already shows it, the replaced unit’s core scanned into remanufacture; recalls execute on these same rails.
“What about counterfeits?” Authenticity by agreement rather than hologram: a scan that resolves to no identity, a serial that appears in two places at once, a custody chain with a gap it cannot explain. The passport either reconciles or it does not, and when it does not, the flag routes to the brand-protection authority with the conflicting records attached.
“What about the circular economy?” The lifecycle does not end at retired. The graph carries material composition for recyclers, recovery and residual value into the lifecycle economics, and a second-life custody chain under the same identity, so reuse, refurbishment, and recycling inherit the same provenance the first life had.
What the capability is
Lifecycle Identity. A persistent identity for every physical asset, lot, and serialized item; every physical marking resolves to it.
Lifecycle Graph. The asset as a sequence of events, created to retired, each with timestamp, source, actor, confidence, and provenance.
Scan & Reconcile. A scan is a truth query: EOS checks every system that claims to know the asset and reports agreement or the exact disagreement, with the named authority.
Lifecycle Value. Acquisition is one line of the real cost. EOS assembles the lifecycle economics executives actually decide on.
Provenance. Never a bare assertion: source, document, clause, last verified, confidence, on every claim.
Agent Context. Agents do not blindly trust one system; they ask EOS to establish state and receive the reconciled, evidence-backed answer.
Contribution without overwrite: manufacturers, suppliers, retailers, and customers can each
contribute their leg of the lifecycle, and nobody rewrites anyone else’s records. EOS reconciles
the contributions, names the authority where they disagree, and routes the decision to the person
who holds it. The action itself is taken in the owning system by its authorized operator; EOS
records the decision and the outcome. Read-only underneath, decision-grade above.
What EOS resolves, and what has to be true before it closes
Every claim on this page follows one rule set. Stating it once, plainly, is
more honest than letting a buyer infer the boundary from six examples.
What applies. Two or more sourced, timestamped claims about the same identified object, where at least one traces to a document, a contract clause, a scan, or a telemetry log. Reconciliation is a fact-versus-fact comparison, not a matter of taste.
What does not apply. Disputes with no system-of-record evidence on either side. Contract ambiguity that requires legal interpretation beyond a named clause’s plain text. Forecasts and opinions, EOS reconciles the inputs to a forecast, never the forecast itself. Anything outside the systems EOS was actually connected to.
Evidence enough to surface. A low bar, deliberately: any single sourced claim that conflicts with another sourced claim. EOS shows the disagreement even at modest confidence, labeled honestly, provenance attached, so the reader judges for themselves rather than EOS deciding what is worth their attention.
Evidence enough to name an authority. A higher bar: a specific document, clause, or scan, not “the system” in the abstract, plus either a pre-established rule for which source wins in this category of conflict, or an explicit named human when no rule exists yet. Confidence is computed from the strength and recency of the evidence, never asserted.
What must be true before a case closes. Four conditions, all four: the authority is named. A decision is made and attributed to that authority, not inferred by EOS. The correction, or the explicit acceptance of the discrepancy, is made in the system of record, and EOS re-checks the record itself rather than taking anyone’s word for it. The outcome is logged with its own evidence.
What happens when it is not resolved. The case stays open. No auto-close on timeout, no most-recent-write-wins, no silence read as agreement. This is the same rule that governs EOS internally: never silently overwrite disagreement.
This is the enterprise-to-physical extension of the same
thesis: reconcile what the ERP says exists against what the warehouse says, what the supplier
says, what the contract says, and what the physical scan says. Run the
reconcile surface, model the cash impact, or
bring us the disagreement.
The commercial structure behind a pilot, four steps from a scoping call to shared value under the Outcome Warranty, is published on the pricing page.