AI agent reads & transcribes

Human decides & validates

Product sheets that are complete, reliable and ready to sell

B2B, industrial and technical catalogues: we produce product data where your sources stop. Packaging, manuals, technical documentation: our AI agent digitises them, our operators verify and validate every sheet before it enters your systems.

Coverage audit 45,876 SKUs · measured by the platform

T0 report

54 %complete
sheets

24,736complete sheets, ready to sell

13,200incomplete sheets: missing attributes or compliance

7,940invisible products: absent from filters and search

Completeness of the fields that power your filters

Dimensions

71 %

Material

48 %

Standards & safety

35 %

Product relations

12 %

Every finding is quantified, exportable and verifiable line by line.

Sample report: measurement first, enrichment second. You decide, numbers in hand.

+15

years supporting enterprises with data and digital services

100 %

of delivered sheets are validated by a human operator

0

fabricated values: unverified data stays empty rather than invented

500

SKUs analysed free of charge before any commitment

The problem

The invisible cost of an incomplete product sheet

On a catalogue of hundreds of thousands of SKUs, incompleteness concentrates in the long tail: accessories, parts, consumables. Every incomplete sheet is a sale that will not happen, and often the add-on sale too.

The direct consequence : a reduced online inventory, a measurable revenue shortfall, and teams busy reworking data instead of selling.

Invisible products

Without complete attributes (dimensions, capacity, material, standard), the SKU appears neither in navigation filters nor in search results. The product is in stock, but the customer never finds it.

Sheets stuck upstream

Empty, partial or unusable supplier files: the SKU waits in a processing queue and revenue starts late across the entire product life cycle.

Non-compliant content

Safety mentions, warnings, manufacturer details, manuals: growing obligations for online sales, which expose the distributor and erode buyer trust.

The goal

Reaching complete product coverage, including SKUs without structured data, historically invisible online.

Our difference

AI digitises, humans validate: never the other way round

PIM, generative AI, supplier catalogues: these tools transform data when it exists. Our work starts where they stop: producing the data present on your products, packaging and manuals but absent from your files, and having it validated by an operator.

AI agent

Reads & transcribes

Analyses your base, detects errors and gaps, transcribes packaging and manuals, and proposes the actions to be taken to the operator.

→

Merkatis operator

Decides, field by field

Reviews each proposal side by side, decides, classifies within your taxonomy and completes what AI cannot produce.

→

Quality lead

Controls & validates

Every sheet is checked against the criteria defined together before delivery: nothing is integrated without human validation.

→

Your PIM / ERP / CMS

Updated, traced

The API writes the validated sheet into the agreed fields. Every update is logged: SKU, fields, date, validator.

No automatic modification, no automatic publication. AI proposes, humans decide.

Our scope

Where the tools stop, our work begins

Many distributors have automated sheet creation from supplier files. These pipelines excel at transforming data when it exists. Three structural limits remain, and they are exactly our scope.

No source data, no sheet

When the supplier file is empty or poor, a generation tool has nothing to transform: it produces nothing, or it invents. We produce the data at the source: packaging, manuals, technical documentation, product photography.

Thousands of suppliers, as many formats

Incoming flows are heterogeneous, incomplete and not aligned with your attribute model. We normalise and structure each SKU according to your taxonomy, category by category.

Generated content is not verified content

On technical or sensitive products (lifting, PPE, chemicals), a wrong characteristic has real consequences. Every sheet delivered by Merkatis is validated by an operator before integration: never automatic publication.

Our positioning : an operational complement to your tools and your teams, not a replacement. Your systems remain the reference, we feed them.

A dedicated platform, not a promise

The platform in action

Three screens, three moments: the AI agent transcribes and proposes, the human decides field by field, the platform measures and proves.

Steer

A traced production queue

Every SKU carries its status, score and languages. Nothing gets lost, everything is tracked: draft, validated, delivered.

Decide

The human decides each sheet

Product photos, OCR workshop, side-by-side FR/NL entry: the operator compares, corrects and validates field by field. AI never classifies, the operator distributes.

Prove

The audit, line by line

Publishability, duplicates, invalid GTINs, language parity: every finding is quantified, exportable, verifiable.

Measure

Is your data ready to sell?

Three scores graded A to E answer three questions: can we work with this base, does this base sell online, is it ready for AI uses. Every component is measured, graded and tracked over time: the completeness rate per family, month after month, not a snapshot at a given moment.

The scope

What we enrich, SKU by SKU

A scope designed for B2B, industrial and technical catalogues, where useful data is first of all accurate data.

Technical attributes: dimensions, load capacities, materials, packaging units, the fields that power your navigation filters.

Product relations: compatible with, accessory of, consumable of, replaces, delivered as structured attributes, the fuel of your cross-selling.

Standards and safety: certifications, warnings, manufacturer details, the mentions expected for online sales.

Titles and descriptions: normalised to your editorial guidelines, consistent formats, no abbreviations or ambiguity, usable for search and SEO.

Product visuals: studio shots, clipping, multiple views, clean visuals for the SKUs that have none.

Associated documents: manuals, technical sheets, safety data sheets, collected, named and attached to the right SKU.

Multilingual translations: validated sheets rolled out to your other markets, with the same verification standard.

Classification: mapping to your taxonomy and completeness of the fields that power your search, category by category.

Our principle: never a fabricated value. Unverified data stays empty rather than invented. Everything is delivered structured to your attribute model, ready to use in your systems and channels.

The method

Four steps, under systematic human control

A turnkey service, from the raw material (the packaging, the manual, the supplier file) to the sheet integrated into your systems. Nothing is asked of your teams.

Collection and digitisation

A Merkatis team handles the collection and digitisation of packaging, manuals and documentation, as well as photo shoots for missing visuals. Our AI agent then extracts the data from this material.

Structuring and harmonisation

Normalisation to your taxonomy and attribute model, category by category. Sheets standardised to the requirements of your channels: e-commerce site, marketplaces, print catalogue.

Enrichment and human control

Technical attributes, regulatory mentions, titles and descriptions to your guidelines, multilingual translations, SEO optimisation. Every sheet is checked and validated by an operator before delivery.

Delivery and integration

Delivered directly into your PIM, ERP or CMS, via API or secure file exchange. A single connection, defined with your IT teams, with no integration fees and no development on your side.

Before any commitment

A free gap analysis, scoped and scheduled

Two families you choose, 500 SKUs, results within 3 weeks, read-only access. An objective measurement, in volumes and per category.

Attribute completeness, measured against your own model, family by family
Product relations that are missing or locked in free text, therefore invisible to your filters
Compliance gaps: missing mentions, warnings and documents, per market
Quantified potential: volume of sheets to enrich, achievable coverage gain, and a pilot proposal billed per validated sheet

Results delivered as a quantified gap report, per category. You decide next, numbers in hand.

Billing

Pay for the validated sheet, nothing else

Deliverables meet the criteria defined together and are only invoiced after your validation.

Connection included

A single API connection, defined with your IT teams, limited to the agreed product scope. No user accounts to create for our operators in your systems.

Non-compliance is never invoiced

A non-compliant sheet is not invoiced: it is reworked. Quality is billable, therefore verifiable.

A report verifiable line by line

Every update is logged: which SKU, which fields, when, validated by whom. The activity report can be checked line by line.

A catalogue is not repaired, it is maintained.

Thousands of suppliers deliver continuously: every incoming flow recreates incompleteness. The longer you wait, the more your coverage degrades. Start with the measurement: the gap analysis is free.

Product Data Enrichment: complete, reliable, ready-to-sell product sheets