How We Integrated AI into All of the Company's Documents: Paperless-ngx at GL.ua

Imagine that you urgently need to find a contract with a client your company worked with five years ago. You remember the client’s last name or the company name, but you don’t know where the document is located: in the accounting department, the paper archive, an email, the ERP system, or on the manager’s computer.

In many companies, documents are still stored exactly this way—in various locations and without any consistent filing system. Some files are named clearly, while others are labeled with names like “scan 145,” “final contract,” or “new latest document.”

As long as there aren’t many documents, the system seems to work. But as the company grows, so does its archive. Searching for a single old document can take tens of minutes and require the involvement of several employees. And if a document can’t be found at all, it’s no longer just an inconvenience—it’s a real risk to the business.

At GL.ua, we decided to consolidate all the documents from our companies, businesses, and startups into a single digital archive. To do this, we deployed Paperless-ngx, connected an office scanner and our ERP system to it, migrated the old archive, configured access rights, and added our own AI assistant.

Translated with DeepL.com (free version)

Why Regular Folders on the Server Are No Longer Enough

A network folder solves only part of the problem: files are stored in one place. But it doesn't address the more complex issues.

• What does the scanned document say?
• Which client or company does it belong to?
• Is it a contract, an invoice, a report, or a shipping document?
• Which employees should have access to it?
• How can you find a document if no one remembers the file name?
• How can you automatically retrieve documents from other corporate systems?

You could create hundreds of folders and write detailed instructions on how to name files. But such a system relies on each employee’s attention to detail. All it takes is one time to put a document in the wrong place or name it incorrectly—and later on, it will be difficult to find the file.

We didn't need just another folder on the server; we needed a full-fledged document management system.

What is Paperless-ngx?

Paperless-ngx is an open-source electronic document management system. It allows you to import documents from various sources, recognize text, store originals, assign document types, tags, and counterparties, manage access rights, and perform full-text searches.

The main difference between Paperless and a conventional file storage system is that the system works not only with the file name but also with its content.

When a scanned document is imported into Paperless, the system performs OCR—optical character recognition. A regular image of a page is converted into a document with a searchable text layer.

After that, you can enter a client’s last name, company name, address, contract number, or any other text—and the system will display the documents in which it appears. Even if the relevant words aren’t in the file name.

A dedicated server for our companies' documents

We deployed Paperless-ngx on a dedicated server within GL.ua's own infrastructure. The system stores documents from our various companies, business units, and startups.

It was important for us to monitor:

• where the documents are physically stored;
• who has access to them;
• how the backup process works;
• how the system interacts with other corporate services;
• what data the artificial intelligence can access.

We’ve created users, groups, and specific roles. The accounting department works with its own documents, managers have the level of access they need, and other employees see only the materials necessary for their work.

In other words, Paperless has become a shared corporate archive, but not a general folder open to everyone without restrictions.

How We Migrated the Old Archive

The first major task was to transfer the documents that had accumulated over the years.

We uploaded the electronic files directly to Paperless. We began gradually digitizing the paper documents. At the same time, we didn’t want to create a new, complicated procedure for employees to follow when saving each document.

So we connected the office scanner directly to the system.

Now, an employee simply places a document in the scanner and presses a button. They don’t need to save the file to their computer, come up with a name, create a folder, and then manually upload the document to the archive.

The scan is automatically sent to Paperless. The system determines the page orientation, aligns the pages, prepares the image for recognition, and performs OCR. As a result, the archive receives not just a photograph of the paper, but a full-text document that can be searched by content.

Integration of Paperless with our ERP system

Simply digitizing old documents isn’t enough. If new invoices and reports still have to be entered manually, within a few months the company will once again end up with several parallel archives.

That’s why the next step was to integrate Paperless with our ERP system.

Now, invoices, reports, and other documents generated in the ERP are automatically sent to the electronic archive. Employees no longer need to download a file from one system just to upload it to another.

As a result, Paperless has become the single point of document storage, regardless of their source. A document can come from a scanner, the ERP system, email, an employee’s computer, or via our AI assistant.
 

Automatic Document Classification

One of the most interesting features of Paperless-ngx is the automatic assignment of document types, counterparties, tags, and storage paths.

Initially, a user helps the system understand the company’s documents. For example, they manually specify that a certain file is a contract, another is an invoice, and a third is a report. They also add the counterparty and the necessary tags.

Once the system has enough correctly categorized examples, it begins to recognize patterns. New documents can then automatically be assigned the appropriate type, counterparty, tags, and storage location.

This doesn’t mean that after the first invoice is uploaded, Paperless will flawlessly organize the entire archive. Automatic classification requires correct examples, and important documents still need to be reviewed by a human.

However, as the number of organized documents increases, a significant portion of the routine work is performed automatically. It is in this way that machine learning brings practical benefits to businesses.

Full-text search instead of browsing through hundreds of folders

Now, to find all documents related to a specific client, all we need to do is enter their last name or company name.

Paperless searches not only titles and details but also the recognized text within the documents. The results show contracts, invoices, reports, and other files that mention the client in question.

Additionally, documents can be filtered by date, type, counterparty, tags, user, and other parameters.

A search that used to take half an hour and involve several people is now completed in a matter of seconds.

Your Own AI Assistant for Working with Documents

After creating a well-organized digital archive, we added another layer of automation—we integrated our own office AI assistant into Paperless.

Employees can interact with it via Telegram.

If an employee receives a document in the messenger, they don’t need to first download the file, log into Paperless, and manually transfer it to the archive. The document can be sent to the AI assistant, which will then forward it to the system for further processing.

You can also ask the secretary to find documents written in plain language. For example:

• Find all records related to a specific company from the past year;
• Display the contract with the desired client;
• Find documents that mention a specific address;
• Analyze the contract and briefly summarize its main terms;
• Compare information across several documents.

An AI assistant can search for, read, and analyze documents, but it must take into account the established access rights. An employee should not receive materials through artificial intelligence that they are not authorized to view within the system itself.

This is an important part of proper corporate AI integration: artificial intelligence should not become a way to circumvent security rules.

What Our Business Has Gained

As a result, we have created not just an electronic archive, but a unified document management system.

Documents are automatically imported from various sources, recognized, indexed, classified, and made available for full-text search. The old paper archive is gradually being transformed into a well-organized database that can be accessed by people, an ERP system, and artificial intelligence.

For businesses, this means:

• less manual file transfer;
• faster search for the information you need;
• less reliance on file names and folder structures;
• a single location for storing documents;
• controlled access rights;
• automatic import of documents from other systems;
• the ability to integrate AI for search and analysis;
• preservation of originals and creation of backups in accordance with company policies.

That said, automation can be implemented gradually. It’s not necessary to migrate the entire archive or integrate all systems right away. You can start with the largest flow of documents of a single type—for example, invoices and reports—and then add new sources and processes.

How to Implement Paperless-ngx in Your Company

Installing Paperless-ngx is just the first step. For the system to truly benefit your business, you need to plan ahead:

• the structure of document types, tags, and counterparties;
• user roles and access rights;
• the process of migrating the old archive;
• connecting scanners and email;
• integration with ERP, CRM, or accounting systems;
• backup and recovery;
• rules for AI access to corporate data;
• human review of mission-critical documents.

We’ve implemented this system for our own companies, so we know it firsthand—not just from the documentation. The GL.ua team can deploy Paperless-ngx on a dedicated server, migrate your existing archive, and configure users, roles, scanning, backups, and integration with corporate systems.

We can also integrate artificial intelligence to add, search for, read, and analyze documents in accordance with your company’s processes and security policies.

Want to create a unified electronic archive and integrate AI into it? Contact GL.ua:

• website: https://gl.ua;
• email: [email protected];
• freephone number: 0 800 759 759.

We can help you transform scattered files and paper documents into a well-organised system where you can find the information you need in a matter of seconds.