Many companies work with large amounts of product-related information every day. Technical datasheets, safety documents, certificates, bills of materials and other files often contain important information about individual products.
The problem is that this information is usually spread across many different documents and formats. Finding specific details, comparing information and transferring everything into a structured system can take considerable time.
This is where artificial intelligence could become useful.
The Problem With Unstructured Product Data
PDF documents are still one of the most common ways to store and share technical information. They are convenient for people to read, but not always easy for software systems to process.
Imagine a manufacturer managing hundreds of products. Each product might have several documents containing information about materials, components, technical specifications or suppliers.
If someone needs a specific piece of information, they may have to open multiple files, search through different pages and manually transfer the relevant details.
This process can become difficult to manage as the number of products and documents increases.
How AI Can Help
AI-based document processing offers one possible approach.
Instead of manually reviewing every document, AI systems can help identify relevant information and extract it into structured formats.
For example, information from different PDF files could be organized into categories such as:
- Product identification
- Materials and components
- Technical specifications
- Supplier information
- Safety-related information
- Certificates and supporting documents
The extracted information can then be reviewed and transferred into databases or other systems.
This does not mean that every document can be processed perfectly or that human review is no longer necessary. Document layouts, missing information and inconsistent terminology can still create challenges.
Why Data Validation Matters
Extracting information is only one part of the process.
It is equally important to understand whether the extracted data is complete, consistent and supported by the original documents.
For example, an AI system might identify a material in a technical document but fail to find the corresponding information in another file.
Such cases should not simply be ignored. Missing or inconsistent information needs to be identified and reviewed.
This is particularly relevant when product information is used for regulatory documentation or other processes where accuracy matters.
AI can support this work, but responsibility for reviewing and validating information remains important.
The Connection to Digital Product Passports
One area where structured product data is becoming increasingly relevant is the Digital Product Passport (DPP).
The European Union is introducing requirements for digital product information under its Ecodesign for Sustainable Products Regulation (ESPR), with additional requirements and timelines depending on the product category.
A Digital Product Passport is intended to make relevant product information accessible in a digital format.
However, before information can be presented digitally, companies first need to collect, organize and maintain the underlying data.
This is where document processing, data extraction and validation could play a useful role.
The challenge is not simply generating a QR code. The information behind it needs to be organized and maintained as well.
A Practical Perspective
I became interested in this topic while working on DPP-Flash, a project focused on simplifying parts of the Digital Product Passport preparation process.
One thing that became particularly interesting to me was how much relevant product information is already available inside companies, but not necessarily in a format that can easily be reused.
Many of the challenges are not necessarily about creating completely new information. They are also about finding existing information, bringing it together and making it usable.
I think this is an interesting area for further development, especially as AI-based document processing continues to evolve.
Final Thoughts
Turning hundreds of PDF documents into structured product data is not just a matter of uploading files and letting AI do everything.
It involves several steps: identifying relevant information, extracting it, organizing it, checking its consistency and making sure it can be used in the right context.
AI can potentially reduce some of the manual work involved, but reliable data still requires proper validation.
As more industries move towards digital product information, finding practical ways to manage existing data will likely become an increasingly important topic.
Disclosure: I am part of the team behind DPP-Flash, a project working in the area of Digital Product Passports.