Digital product passport
One QR code printed on the label turns a pile of scattered PDFs into a page that answers whoever is holding the product, for as long as that product exists: what it is, how to use it safely, where the current documents are, whether the batch has been recalled and where to buy more.
No card required. We import your first product live, from your own files.
The problem
You sell through distribution, so you do not know who is using your product, where it sits or which revision of the manual they are reading. There is no return channel — right up to the day you urgently need one.
You find a defect in a batch produced eight months ago. You know which wholesalers took it. You have no idea which shelves it is on today.
A recall ends as a notice on your own website and a hope that the distributor passes it along. The person actually holding the product never hears about it.
A banner on every scan, plus push and email to holders — with a live counter of how many have seen it.
The manual in the box is from the day it was printed, not the day it is used. A changed standard means reprinting labels or living with the discrepancy.
Blocks hold a reference to the document, not a copy of the file. One new revision updates the whole catalogue at once.
Your customer runs out of material and types the name into a search engine. They land anywhere but with you — while holding your packaging in their hand.
Ordering straight from the product page, no account and no storefront funnel. The shortest path from "running out" to "ordered".
GPSR requires responsible-entity data with the product. The digital product passport becomes mandatory category by category — batteries from February 2027. Medical devices need UDI per SKU.
Compliance blocks cannot be deleted and they block publishing until they are complete.
How it works
The manufacturer describes the product once, and the information reaches everyone holding it — with no distributor in between and no label reprint every time the documentation changes.
A PDF manual, a safety data sheet, a photo of the label, a link, a video. The import pulls the content out of them and assembles a finished product page using your category template.
Generate a pool of codes at model, batch or individual item level. The export goes wherever you actually print.
From that moment the code is your channel. Change a document and it changes for everyone. Recall a batch and holders see it on their next scan.
Features and benefits
This is not a capability list. Each part of the system exists because something in the product lifecycle was impossible or expensive. Below: what it does, and what you get from it.
The canonical Itemoo token comes at three levels: model, batch or single item. It is random, so it reveals neither your numbering nor your production volume.
Short production runs and prototypes get a digital identity the same day, without waiting for a GS1 number pool.
Already have a GTIN? The same code serves both the checkout and the product page, in the /01/{gtin}/10/{lot}/21/{serial} format, with the check digit validated on save.
One label instead of two, and Sunrise 2027 readiness without another implementation project.
The operator scans the first and last code on the sheet, and the entire pool between them is assigned to the batch automatically.
Nobody types a thousand numbers, so there is no typo sending a customer to somebody else's product page.
PDFs, HTML, label photos and video go in together. The system detects the category, fills the blocks and returns a reading report: what it read, how many pages, and what could not be parsed.
Describing a typical product drops from hours of catalogue-team writing to minutes of review.
The data contract requires every model-written statement to cite a source document and a quotation. Missing evidence is a validation error, not a more cautious sentence on the page.
You can publish safety-relevant content without re-reading every paragraph in fear of an invented figure.
Every field can carry versions for many languages, and every block its own market list. Safety data sheets, passports and legal documents never pass through machine translation.
You enter new markets without the risk that a translation quietly changes wording you are legally answerable for.
Announce it for a batch or a LOT range. A red banner appears on every scan with the reason and return instructions, and a counter shows how many holders have already seen it.
You hold proof of reach for the regulator, instead of a statement that a notice was on your website.
Blocks hold a document reference, not a copied file URL. A version can carry a future effective date and the system switches to it on its own. A "where used" view shows the blast radius before you act.
One new manual revision updates the whole catalogue, while each batch keeps the version valid on its production date as evidence.
A product leaves the range but its printed code keeps working — it shows a withdrawn-from-sale notice and a link to the successor. The panel shows what blocks the action before you click.
A five-year-old code never lands on an empty page, so your archive never becomes a source of complaints.
Your customer is holding an empty bottle standing next to the machine. They order from that same page, the cart survives the trip to the payment gateway, and the address stays on their device.
Forced registration disappears — and it accounts for roughly a quarter of abandoned carts.
Two-way filtering across up to four axes, search by dimension and index prefix, and pairing completed from the manufacturer's own matching table. Before the configurator: "your most frequent" and "recently ordered".
A technician reaches the correct variant without phoning sales, and you stop processing returns caused by a wrong index.
The brand app installs straight from the product page. Push and email go out only for recalls and new document revisions.
The channel keeps its credibility — a false alarm teaches people to ignore every alarm, including the one that counts.
Signal analysis from data you already collect: the same serial number in three countries, a country change within six hours, four hundred scans of a single item. Every signal ships with its legitimate explanation.
You find the counterfeit channel without holograms, taggants or any change to your production process.
An entry point at /.well-known, product data as JSON, and schema.org JSON-LD injected into the page. Every fact points to its source document and that document's version.
When an assistant picks between ten descriptions of the same resin, manufacturer-signed data with a cited document wins.
Scans by country and time, revenue, catalogue completeness, and the list of questions the assistant could not answer from your material.
Unanswered questions are a ready-made list of documentation gaps, collected from people using the product right now.
What sets us apart
These are not slide-deck slogans. Each one is enforced by the data contract or the architecture, so it cannot be waived when a deadline gets tight.
Every model-written statement must cite a source document and a quotation. The schema demands it (evidence: min(1)) — not a line in a prompt. A hallucination ends as a validation error before anyone sees it.
This single rule shapes the rest of the system: archiving instead of deletion, document versioning instead of file replacement, and no brand identifier change without a redirect. A label outlives a commercial decision.
Safety data sheets, product passports and legal documents run in cite-only mode. The model does not paraphrase them, does not translate them and does not invent file addresses. The forbidden-field list is hard and covered by tests.
An exhausted AI allowance switches off import and the assistant. It never switches off recalls, the product page or the documents. User safety is not a line item in a pricing table.
With counterfeit signals we show the data alongside the legitimate explanation that could account for it. Whether it is a clone or a wholesaler with three warehouses is decided by the person who knows the distribution channel.
Proof, not claims
The values below are design targets written into the system's non-functional requirements and enforced in code — not an averaged marketing statistic.
| Path | Target | Why that number |
|---|---|---|
| Code resolution and recall banner | 99.9% | A printed code has to work |
| Shop and checkout | 99.5% | No sales, not no information |
| Manufacturer panel | 99.5% | The work can wait |
| Product assistant | 99.0% | The question reaches you either way |
| AI import | 98.0% | A deferrable operation |
| Failure | Still works |
|---|---|
| AI model provider | Product page, recalls, documents, shop. The question form still accepts the question and forwards it to you. |
| Payment gateway | Product page, documents, cart preserved for the return trip |
| Mail server | Everything except email; the content goes to the log |
| Push notifications | Email as the fallback channel plus the scan banner |
| File storage | Product page with its full text content |
| Customer has no connection | The page from device cache after the first visit |
| AI quota exhausted | Recalls, product page and documents — always |
GPSR: you enter responsible-entity data once and it appears with every product, and missing data blocks the first publication. DPP: the passport block covers materials, origin, carbon footprint and recycling — batteries from February 2027. UDI: GTIN per variant, because medical devices need a separate identifier for each SKU.
A scan records the country, not a precise location, and is never tied to a person. A shipping address typed before purchase stays on the customer's device. Customer accounts are passwordless — sign-in by emailed code, so there are no passwords to leak. Contact data lives in a separate table so it can be permanently erased.
Startup halts on unsafe configuration rather than quietly shipping a hole. Files are validated by content, not by extension. The language model has no tools and no write access — everything it reads is data to it, never an instruction.
Who it is for
The more documentation travels with your product, and the longer that product lives with the customer, the bigger the difference. Category templates are configuration rather than code — a new industry is a new row, not a new project.
Catalogue teamDescribes hundreds of indexes. Gets import from source material, CSV column mapping and a gap list they can tick off — instead of an empty form.
Shop floor operatorPrints and activates codes by range, in gloves, with no access to prices or settlements. Serial numbers arrive straight from MES.
Quality departmentUploads a new document version with a future effective date and sees the blast radius before acting. Announces recalls without queueing for IT.
Sales teamRuns the fulfilment queue with statuses and sees the questions customers ask at the product — real demand and real gaps in the description.
Technician and labScans, picks the right variant, downloads a parameter profile for their own equipment and orders without creating an account.
Regulator and auditorGets proof that a recall reached holders, the document version history, and which version was in force on a given batch's production date.
Testimonials
We have no rollouts we could describe in someone else’s words yet — and we would rather say so than fill this section with invented quotes.
An editorial note. Itemoo is ahead of its first public deployments, so we do not print invented testimonials or logos of companies that do not use us. Until real quotes exist, the evidence is the verifiable numbers in Proof and a demo run on your own material.
Plans
Every plan shares the same core: codes, the product page, documents and recalls. They differ in how much AI work you do and how fast we respond to a ticket.
A first catalogue, one product line, testing the idea on your own products.
A full production catalogue, several markets, catalogue and production teams working in parallel.
Regulated industries: medical devices, chemicals, batteries. Audits, oversight and legal deadlines.
Straight talk about payments. Itemoo is not a payment institution and does not handle your money. Every manufacturer operates their own merchant account at the payment gateway — sales proceeds go from the gateway directly to you, bypassing us entirely.
FAQ
If yours is not here, write to us — we answer concretely, including when the answer is "we do not do that".
A digital product passport (DPP) is a set of data about a product — materials, origin, carbon footprint, recycling route — reachable through an identifier carried on the product itself. The EU is phasing it in by category: batteries from February 2027, further groups in the years after.
If you make batteries, textiles, electronics or furniture, the obligation either already applies to you or will within a few years. Itemoo ships the passport block as standard so you can fill it in before the deadline becomes a problem.
No. The canonical Itemoo code works independently of GS1 and suits short production runs, prototypes and products never sold at retail.
If you do have a GTIN, we attach GS1 Digital Link as a second address for the same product — so one code serves both the retail checkout and the product page, in line with the Sunrise 2027 direction.
Two routes, usually in parallel. Import from source material takes manuals, safety data sheets and label photos and builds finished pages — a typical product means minutes of review instead of hours of writing.
CSV import takes an ERP or PIM export and proposes a column mapping that you approve by hand. The model creates no content here, it only matches column names to fields — a deliberately narrow and safe role.
The data contract does not allow it. Every fragment written by the model must cite a source document and a quotation — missing evidence means the fragment is rejected at validation, rather than turning into a more cautiously worded sentence on the page.
On top of that, safety data sheets, product passports and legal documents run in cite-only mode: the model does not paraphrase, translate or generate file addresses for them. A control set of 34 cases and 330 assertions must pass in full before every release.
You announce the recall for a specific batch or a LOT range. From that moment every scan of a code from that batch shows a red banner with the reason and what to do — before the user sees anything else.
In parallel a push notification goes to installed brand apps, with email as the fallback. A counter shows in real time how many holders have seen the notice — that is your proof of reach for the regulator. Recalls are never subject to plan quotas.
No, and that is the heart of the idea. The code on the label points at the product, and the product page points at the document by reference rather than by a copied file URL. Uploading a new version changes the documentation across the whole catalogue at once.
A version can carry a future effective date and the system switches to it on the right day by itself. Batches produced earlier keep the version that was in force on their production date, because that is the evidence for that specific production run.
From the data we already collect on every scan: the code, the time and the country. The same serial number cannot be in Gdańsk and Guangzhou at once, and a genuine item is not scanned four hundred times.
The system surfaces signals — multiple countries, an impossible journey, unusual volume — always next to the legitimate explanation that could account for them. We do not issue a verdict, because you are the one who knows your distribution channel. No new technology, no change to production.
You enter responsible-entity data once when registering the brand, and it appears with every published product. Gaps in critical data block the publication of your first product — the system will not let a non-compliant page go out.
We split responsibility plainly: Itemoo is answerable for making sure content unsupported by your material never reaches the page. The manufacturer is answerable for what is in the material, and for the decision to publish.
No. A scan records the code, the time and the country — not a precise location, and with no personal identifier. A shipping address typed before purchase stays on the customer's device and reaches us only when they place the order, with consent.
Customer accounts are passwordless: sign-in by a code sent to email. The risk of a password leak disappears because there is nothing to steal. Accounts belong to the brand, not to Itemoo, and contact data lives in a separate table so it can be permanently erased on request.
In two layers. The first is a tenant identifier filter on every application query. The second is enforced row-level access rules inside PostgreSQL itself, active across fourteen tables, with a dedicated database role for the application.
The second layer exists precisely so that a bug in the first cannot expose someone else's data. For regulated industries a hybrid option is available: source material and production data stay with you, and only published pages and public files live in the cloud.
No. A product enters the brand shop automatically once it is published and priced — you do not maintain a second catalogue alongside the one you already have.
Customers buy without creating an account, because forced registration accounts for roughly a quarter of abandoned carts. We offer the account after payment, and a guest order can be found using its number and email address.
You do not need IT to reach your first working code: register the brand, import the material for your first product, print a sheet of codes on an ordinary printer. That fits into a single afternoon.
IT gets involved only for the integrations you want anyway: serial numbers from MES, the catalogue from ERP, Zebra printers on the line, pointing your own domain at the pages. Each of these runs through a ready import or export, with no code to write on your side.
For the demo we take your manual and your safety data sheet, import them live, and show you the finished product page together with a code you can scan immediately. Thirty minutes, no slide deck.
No card, no commitment, no fifteen-email sales funnel.