Question 1

What is Docxi?

Accepted Answer

Docxi is an AI-native product-data onboarding platform that transforms messy supplier documents—PDFs, datasheets, spreadsheets, and mixed catalogs—into structured, validated, and publish-ready product data. We sit upstream of your PIM, ERP, or ecommerce platform and handle the complex work of extraction, classification, and validation before the data ever reaches your systems.

Question 2

Who is Docxi for?

Accepted Answer

We work best with mid-sized distributors, manufacturers, retailers, and marketplaces that receive supplier catalogs regularly, manage several thousand to 100,000+ SKUs, handle multiple suppliers with different file formats, and have a catalog or product-data team that spends significant time on manual data preparation.

Question 3

Is this a PIM replacement?

Accepted Answer

No. Docxi is not a PIM. We are an upstream intake and quality layer that prepares supplier data before it enters your PIM, ERP, or other downstream systems. Think of us as the bridge between messy supplier content and your structured product database.

Question 4

How is this different from PIM AI features?

Accepted Answer

Most PIM AI features assume the data is already reasonably structured and focus on enrichment inside the PIM. We specialize in the messy reality: multi-product catalogs, multi-page datasheets, mixed ZIP packages, and multiple schemas—with field-level evidence, human review, and PIM-agnostic export.

Question 5

What types of files can you process?

Accepted Answer

We process PDFs (text-based and scanned), datasheets, product catalogs, spreadsheets (Excel, CSV), ZIP packages with multiple files, mixed document packages, large catalogs (500+ pages), and composite files with multiple product types.

Question 6

Can you handle one file with multiple product types?

Accepted Answer

Yes. If a single PDF or ZIP contains multiple product families, we detect logical product groups, classify each against your taxonomy, assign the appropriate schema, extract each independently, and present them for review separately.

Question 7

What if one product spans multiple pages?

Accepted Answer

We correlate product data across pages using shared identifiers like MPN, SKU, product name, and manufacturer. All attributes from different pages are merged into one canonical product record, with each field linked back to its source page for traceability.

Question 8

Do you support multiple taxonomies and schemas?

Accepted Answer

Yes. We support multi-level taxonomy hierarchies, multiple schemas for different product types, supplier-specific templates, schema versioning, schema inheritance, and conditional and optional attributes.

Question 9

How do you handle data validation?

Accepted Answer

We validate at two levels. Product-level: required fields, valid data types, units and normalization, allowed values, business rules, and duplicate detection. Package-level: duplicate products across files, conflicting taxonomy assignments, missing product groups, and unclassified content.

Question 10

What happens if extraction is uncertain?

Accepted Answer

Low-confidence fields, conflicts, or missing attributes are routed to a human review queue. Your team reviews only the exceptions, not everything. This keeps manual effort low while maintaining quality.

Question 11

Do you preserve source evidence?

Accepted Answer

Yes. Every extracted field retains provenance: source file name, page number, bounding box or chunk reference, extraction confidence score, and schema version used. This is critical for auditability, compliance, and troubleshooting.

Question 12

Can you integrate with our existing PIM or ERP?

Accepted Answer

Yes. We export structured data in CSV, Excel, JSON, XML, API-based exports, and direct integration via webhooks or connectors. We are PIM-agnostic by design and can work with multiple downstream systems.

Question 13

How does the workflow work?

Accepted Answer

The typical workflow is: Upload supplier files, system identifies the supplier and detects product groups, products are classified into taxonomy and assigned schemas, AI extracts attributes with page-level evidence, your team reviews exceptions, and approved data exports to your PIM, ERP, or ecommerce platform.

Question 14

Do we need to configure everything manually?

Accepted Answer

No. We support automatic supplier identification, automatic taxonomy classification, template reuse for repeat suppliers, schema inheritance from parent categories, and default validation rules. Manual configuration is only needed for exceptions, custom schemas, or complex business rules.

Question 15

Can we reuse templates for repeat suppliers?

Accepted Answer

Yes. Once a supplier's format and schema are mapped, the system reuses the template for future submissions. This significantly reduces recurring onboarding effort.

Question 16

How do you handle updates from suppliers?

Accepted Answer

When a supplier sends an updated catalog, we identify changed, added, or removed products, extract only the differences, flag uncertain changes for review, and preserve historical versions for comparison. This avoids reprocessing everything from scratch.

Question 17

Can multiple team members work on the same intake?

Accepted Answer

Yes. We support role-based task assignment, multi-user review workflows, approval chains, audit logs, and comments and notes on specific fields or products.

Question 18

What if we have different approval processes for different categories?

Accepted Answer

We support configurable workflows: category-specific approval rules, different reviewers for different product types, conditional routing based on schema or taxonomy, and partial approval so clean products can proceed while others are under review.

Question 19

What is the catalog health review?

Accepted Answer

The catalog health review is a free, consultative assessment of your current supplier onboarding process. We review a sample supplier catalog, identify manual effort and quality risks, estimate onboarding time and error rates, provide a short improvement report, and discuss whether a pilot makes sense.

Question 20

Is the catalog health review really free?

Accepted Answer

Yes. It is a consultative service to understand your situation and see if our approach can help. If it does not make sense for you, we will tell you honestly.

Question 21

What is involved in a pilot?

Accepted Answer

A typical pilot includes one customer, one product category, one or two suppliers, one real catalog or document package, 2–4 weeks duration, and a measurable before-and-after comparison. We work with you to define success criteria upfront.

Question 22

How is pricing structured?

Accepted Answer

Pricing typically includes an implementation or setup fee (one-time), a monthly platform fee, a usage fee based on processed product records or pages, and an optional managed catalog service. We avoid pricing purely by AI calls.

Question 23

Do you offer enterprise pricing?

Accepted Answer

Yes. For larger organizations with high volume, multiple business units, or complex integration needs, we offer custom enterprise pricing. Contact us to discuss your requirements.

Question 24

Where is our data stored?

Accepted Answer

We use secure, enterprise-grade cloud infrastructure with encryption at rest and in transit. Specific data residency options are available for enterprise customers.

Question 25

Is our supplier data confidential?

Accepted Answer

Yes. We treat all supplier documents and extracted data as confidential. We can sign NDAs and have standard data protection agreements in place.

Question 26

Do you comply with data protection regulations?

Accepted Answer

We follow industry best practices for data protection and can support compliance requirements such as GDPR, CCPA, and other regional regulations. Contact us for specific compliance questions.

Question 27

Can you delete our data?

Accepted Answer

Yes. You can request deletion of your data at any time. We have standard data retention and deletion policies that can be customized for enterprise customers.

Question 28

How long does implementation take?

Accepted Answer

For a standard pilot, implementation typically takes 1–2 weeks. For full production deployment, timeline depends on taxonomy complexity, number of suppliers, integration requirements, and custom workflow needs. We provide a detailed timeline after the catalog health review.

Question 29

Do you provide training?

Accepted Answer

Yes. We provide initial onboarding for your catalog team, documentation and user guides, ongoing support during the pilot, and regular check-ins for enterprise customers.

Question 30

What kind of support do you offer?

Accepted Answer

Support options include email support, scheduled calls for complex issues, a dedicated support channel for enterprise customers, and regular business reviews for strategic accounts.

Question 31

Can you help us clean up existing product data?

Accepted Answer

Yes. In addition to new supplier onboarding, we can help with legacy catalog cleanup, data migration to new PIM or ERP, schema remapping and normalization, duplicate detection and resolution, and attribute standardization.

Question 32

What AI models do you use?

Accepted Answer

We use a combination of layout detection models, large multimodal models for extraction, classification models for taxonomy assignment, and custom validation rules. Our pipeline is designed for reliability, traceability, and continuous improvement.

Question 33

Do you support OCR for scanned documents?

Accepted Answer

Yes. We support OCR for scanned PDFs and image-based documents using modern OCR engines combined with layout-aware extraction to handle tables, multi-column layouts, and mixed content.

Question 34

Can you handle non-English documents?

Accepted Answer

We support multiple languages for extraction. Language support depends on the specific document and product type. Contact us if you have specific language requirements.

Question 35

Do you provide APIs?

Accepted Answer

Yes. We provide REST APIs for file upload, status checking, data export, and integration with downstream systems. API documentation is available for technical teams.

Question 36

Can you process images and diagrams?

Accepted Answer

Yes. We can extract information from product images with embedded text, technical diagrams, charts and graphs, and tables and specifications in image format. We use vision models to decode visual information.

Question 37

How do we get started?

Accepted Answer

The easiest way to start is to request a catalog health review. We schedule a brief call, review a sample supplier catalog, provide a short improvement report, and discuss whether a pilot makes sense.

Question 38

What do you need from us to start?

Accepted Answer

To begin a catalog health review, we need a sample supplier catalog or datasheet, information about your current onboarding process, your target system (PIM, ERP, ecommerce), and any existing taxonomy or schema documentation if available.

Question 39

How quickly can we see results?

Accepted Answer

For a pilot, you can see results within 2–4 weeks. The timeline depends on document complexity, number of product types, your review and approval process, and integration requirements.

Question 40

Who should we contact for more information?

Accepted Answer

You can reach us through the contact form on this website or email us directly. We typically respond within 1–2 business days.

Question 41

Will this replace our catalog team?

Accepted Answer

No. The goal is to reduce manual, repetitive work so your team can focus on higher-value activities like quality improvement, supplier collaboration, and strategic initiatives. Human review remains an important part of the workflow.

Question 42

What if our suppliers send very inconsistent data?

Accepted Answer

That is exactly the problem we are designed to solve. We handle different file formats, inconsistent attribute naming, missing or incomplete data, mixed product types, and multi-page documents.

Question 43

Is this just another AI tool that will create more work?

Accepted Answer

No. We are designed to reduce work, not add to it. We automate repetitive extraction, route only exceptions to human review, preserve evidence so your team can trust the output, reuse templates for repeat suppliers, and integrate cleanly with your existing systems.

Question 44

What if we already have a PIM with AI features?

Accepted Answer

Many of our customers have PIMs with AI features. We complement them by handling the upstream complexity that PIM AI often does not address well: complex supplier documents, multi-product packages, cross-page correlation, field-level evidence, and multi-schema routing.

Question 45

How do you measure success?

Accepted Answer

We measure success by reduction in manual touch time, faster onboarding cycles, fewer missing attributes, lower error rates, higher acceptance rates without correction, and time from file receipt to publish-ready data. We define specific metrics for your pilot.

Turn messy supplier catalogs into trusted product data

Your PIM can store product data. But getting it in is still painful.

Inconsistent supplier files

Mixed product types in one catalog

Details scattered across pages

Manual, repetitive work

Delays that cost launches

An intake layer designed for complex supplier documents

Multi-page PDFs and large catalogs

Multiple schemas in one package

Cross-page product correlation

Supplier templates and taxonomy mapping

Field-level evidence and traceability

Human review and exception workflow

Designed around your workflow, not the technology

Upload

Identify

Classify

Extract

Review

Publish

Why this is not just another PIM AI feature

Built for teams drowning in supplier spreadsheets

Start with a catalog health review

Frequently asked questions

Ready to reduce manual catalog work?