TAdviser: "EDMS and ECM Day 2025"
On September 18 To Moscow TAdviser , he held a conference on electronic document management systems and content management platforms (ECM). Speakers assessed market trends, shared experience in introducing systems, presented vendor solutions, talked about the use of artificial intelligence for registration, processing and analysis of documents.
Content |
The conference was attended by representatives of such organizations as Gazprom Neft, the Bank of Russia, the Fund for the Development of Effective Cybersecurity, Rostelecom, the Kurchatov Institute Research Center, Tasty - and Tochka, Dixy-South, International Bank for Economic Cooperation, Sberbank Leasing and many others. Olga Rostunova, director of the Mechel department, moderated the events.
Hundreds of thousands of documents a month
Victoria Bondareva, Deputy General Director, NPF Future, spoke about her experience in using artificial intelligence when processing incoming documents. AI here decided to implement for many reasons. The decision to combine funds and the need to ensure the registration of documents in accordance with the requirements of the law within the established time frame and a large volume of documents for the group of funds influenced. In total, 337 thousand such documents were collected in 2024 alone.
After studying the ready-made systems market, it was decided to jointly develop its own system with the vendor. The partner vendor was presented with key requirements for the future product. I would like the latter to be able to recognize a structured and unstructured document, and that the quality of recognition is at the level of 90% for printed text and 60% for handwriting. Import independence was also required.
The joint project started in September 2024. It is now complete. The developed module used technologies such as optical character recognition (OCR, Optical Character Recognition), handwriting recognition (HTR, Handwritten Text Recognition), large language model (LLM), machine learning (ML).
 | Before the introduction of artificial intelligence, more than 40% of all incoming documents were processed manually, which led to numerous errors, says Victoria Bondareva. - One of the planned goals of the project was to reduce manual processing by 50%, but as a result, we reduced manual processing by 62%. |  |
The rest of the implementation results look like this: the volume of manual processing of documents and the time of their registration have decreased. The registration quality is improved, the PHY is optimized, and the registration process is functionally divided. For 2026, the company plans to increase the current quality of recognition, as well as the use of AI in other processes (classification of calls, processing of service applications).
Natalia Akulich, Head of the Document Control Automation Center, AgroPromProcurement Group of Companies, began by describing the initial needs for electronic document control. They were formulated in 2021: optimization/unification of document management and accounting processes of legal entities included in the group of companies, ensuring the transparency of document management processes, their reliability and security, simplifying processes when working with counterparties, reducing storage costs, automating work with the archive, ensuring compliance with legislation.
]]
Then the speaker moved on to the stages of the project. First, they created the "Document Flow Automation Center," implemented their own gateway. E-DOCUMENT FLOW Then there was integration with (API "Honest Sign," etc.), implementation based on KEDO /. SabyGSI Automatic document exchange between accounting systems via the Datareon bus was established.
| | At all stages of the project, an analysis of current processes was first carried out, requirements for the future system were formulated. Then it was designed, tested, a pilot project was launched and only then the transition to industrial operation took place, - said Natalya Akulich. | |
Among the results achieved is full compliance with the law - now 170 thousand documents are signed monthly, and the processing time of documents has decreased by more than half. Reduced 40 FTE. Next, the company plans to switch to electronic transport documents. Intelligent pipeline
Alexander Bessarabenko, head of the EDO systems management department, System Electric, stressed that artificial intelligence is just another technology. The first thing to do before implementing it is to audit business processes. "83% of the document flow in our company is in electronic form," he said.
For high-quality document processing, it is important to implement an Intelligent Document Processing (IDP) pipeline, when the document type is recognized, data is extracted from the required fields, checked and sent to the accounting system. Each of these tasks needs its own, correct tool. So, to read the text and analyze the layout of the document, you need to use OCR and computer vision technologies, and to understand the meaning and extract entities - machine learning in combination with NLP.
The speaker considered specific cases of using AI technologies within the company. To check contracts for compliance with internal policies and templates, here you use the LLM and RAG system based on corporate data and checklists. As part of this case, 6 contracts are checked in 100% hours, while about 30% documents with different inconsistencies are identified. Alexander Bessarabenko stressed that the labor costs of lawyers are reduced by 60-70%.
As part of the second case, which the speaker spoke about, the task of extracting serial numbers, dates, employee data from hundreds of acts is being solved. 600 documents are processed in 15 minutes, the accuracy of key fields at the 95% level, the proportion of manual verification is less than 20%.
Dina Shebordaeva, head of the quality and reporting systems department at OSSP, one of the largest road operators in Russia, spoke about the results of updating the EDMS based on DocsVision, as well as plans for its development.
The prerequisites for the project were the need to establish a unified order in the documents of forty-nine divisions of the company for collecting payments. In addition, there was a need to speed up the search for documents and shorten the period for their approval.
As a result of the introduction of a new version of DocsVision, the speed of work in the system has increased - opening a card takes no more than 2 seconds. A complete transition to EDM with counterparties was carried out, interaction with them through the integration of EDM and Kontur Diadoc was simplified, the signing of documents with an electronic signature was implemented, as well as with the help of the MFD.
| | The updated system works exclusively in the web interface, - explains Dina Shebordaeva. - This is very convenient - the work with the system has accelerated significantly, a new full-text search has been carried out, it has become easier to carry out technical support. Even if an employee has not encountered EDMS before, he has no problems finding the necessary information. | |
The EDMS update covered all divisions and 1,300 employees. The plans for the development of the system include integration with the corporate portal Bitriks24"," work on the "Archive Management" module - all taking into account the requirements of the order of Rosarchiv No. 77.
Employees continue to fight back
Another representative of NPF Future, project manager Olga Shevchenko, spoke about the work carried out in the fund with documents in one window through the integration of EDMS and the operator of EDO Kontur.Diadok.
The integration was carried out through a special module, which was subject to certain requirements. In addition to implementing the actual integration with the EDMS and Diadoc through the API, the module had to support the functionality of receiving and signing any formalized and unformalized documents from Kontur. Diadoc "in the EDMS, as well as synchronize the directories of counterparties between the EDMS and" Kontur.Diadok. " A number of other options were needed.
Olga Shevchenko explained how difficult the process of signing a document looked when receiving a document from Kontur.Diadok before automation, when the process was implemented in three systems, including on paper. As a result of the project, eight manual operations were automated. Accounting is excluded from the process of receiving and registering documents. In addition, all documents are now processed and signed in the same system, and the responsible persons directory is implemented.
| | Our project for integration with EDM and EDO is a local, simple case with good efficiency, - Olga Shevchenko summed up. | |
Among the difficulties encountered during the project, the speaker named informal, incorrectly compiled documents from counterparties, which increased manual processing, as well as the fact that not all counterparties use Diadoc.
Difficulties were also associated with resistance to changes on the part of employees. In particular, they did not like the transition to work with packages of documents.
Ekaterina Agarkova, Project Solutions Manager (ECM), BIOCAD, shared her experience of updating the internal ECM system. She listed the prerequisites for updating ECM: vulnerabilities of the solution itself, termination of support for the current Red Hut Enterprise Linux OS, hanging at the PostgreSQL v.13 DBMS level, and a number of other problems.
| | It was necessary to update not only the ECM system itself, but, to a large extent, the IT landscape itself, - shared Ekaterina Agarkova. "That was the difficulty. | |
During the large-scale update, the found problems and vulnerabilities in the ECM were closed, the RHEL 9 operating system was switched, and Java and network components were updated. In addition, a transition was made to the new version of the PostgreSQL v.15 database ("vanilla" version), a full-fledged CI/CD was introduced based on its own configuration collector, and ECM interfaces were finalized in order to facilitate user experience.
The project was implemented by a joint team, with the involvement of external IT contractors. The total number of specialists in the project team is 70 people.
Ekaterina Agarkova gave the following recommendations. It is necessary to automate deployment processes using special tools, go out to update the system with a minimum technical duty and take into account changes in technologies, and carry out regular updates.
Anna Folomeeva, owner of the product, Metro Cash & Carry Russia, recalled that the company introduced Directum RX EDMS back in 2001, and since then the system has been developing in accordance with business needs. She further spoke about the results of three successful projects implemented here on the basis of EDMS: the formation of machine-readable powers of attorney (MCD), integration with the Kontur. Dicont system, which is responsible for automating early payment transactions, and outgoing letters to EDMS Directum RX.
Within the framework of the CDM project, the EDM now automatically generates tasks for creating a power of attorney when a new employee appears with certain responsibilities, and senior managers themselves extend the CDM - the EDM functionality is so simple. In addition, when an employee is dismissed, the MCD recall is automatically triggered.
Anna Folomeeva paid special attention to the success factors of the implemented projects. Thus, the main factors of the success of the Power of Attorney project were constant interaction with the internal customer - the legal division of the company, the development of functionality at the design stage, a high level of training of the external development team.
| | Within the framework of the "Power of Attorney" project, we significantly reduced the time for issuing the MFD - up to 10 minutes. The time spent by the counterparty on signing documents has decreased by 70%, - Anna Folomeeva shared the results. | |
All three projects were implemented by a vendor and a partner - a system integrator. The internal team of the customer remained the correct setting of tasks.
40% of corporate documents will not be agreed by a person
Mikhail Egorov, Senior Executive Director, ID Management Technology, spoke about how artificial intelligence is used to register, route and process documents in EDMS based on the Documento platform, the company's own development.
The speaker presented the product range implemented on the Documento platform: EDMS, archive, SZDO and Digital Chancellery (Central Committee). He spoke in more detail about the Central Committee, which automates the processing of primary documents in the organization, and also presented the program interfaces. The solution works on AI algorithms implemented on the basis of open LLM, which significantly reduce the labor costs of employees. For example, when processing incoming documents, references to other documents are identified, documents are automatically linked.
{{quote 'Digital Chancellery' is a separate system that automates classification, attribution, information enrichment, analysis, routing of the entire incoming flow of documents, - added Mikhail Egorov. - The system is autonomous, it is built into any IT landscape and does not even require the conversion of the basic EDMS. }}
Saving time for processing documents, increasing the speed of registration, accuracy and quality of filling fields in the document card, the possibility of unlimited extension of attributes, summarizing documents - these are the general effects of the EDMS implementation. The system can be deployed at the customer's site or provided as a cloud service, said Mikhail Egorov. He stressed that the Documento platform is included in the Register of Russian software, compatible with Russian browsers and OS, batch licensing, depending on the number of incoming documents.
| | Where does the need for the use of artificial intelligence in EDMS come from? - the speaker asked. - Systems are becoming more complex, there is more and more information in them, so it is more difficult to train users. Convenience and flexibility of user interaction with the system come first. | |
In the EDMS THESIS, a combination of robots and neural networks was created using AI, it is possible to use AI agents to interact between systems, and everything is implemented by system settings, and not coding, the speaker noted.
EDMS allows you to use any LLM models and in any number. The formats of artificial intelligence technologies in the EDMS THESIS are:
- AI assistant - convenient interaction with AI through a text field, flexible interaction setup through a reference book, chat bot.
- AI module - all the functionality of the assistant, supplemented by monitoring and collecting information about users, robot designer, LLM connection in any tasks.
- Interactive simulator - solution at the junction of DAP (Digital Adoption Platform) and LMS (Learning Management System) technologies - reduced time to adapt a new employee.
Among the implementation effects, Alexander Vorobyov named savings in development, optimization of time and costs, improvement of processes in the system and a number of others.
Igor Kaibanov, analyst, expert at EDMS/BPMS, Axenix, at the beginning of his report outlined the historical evolution of EDMS from electronic archives to systems in convergence with ECM and CMS. The speaker is confident: SEDs begin to take work from people and evolve into predictive assistants. According to Gartner's forecast, by 2027, up to 40% of corporate documents will be created, agreed and archived without human input, he recalled.
| | The document today becomes a managed asset requiring economic assessment, the speaker believes. - The main problem when working with documents is the human factor, which has such negative aspects as incompetence and burnout. | |
Igor Kaybanov listed trends in the development of modern EDMS. Among them are the formation of a dynamic context, platform AI solutions that allow you to collect specialized assistants and autonomous background agents, end-to-end integration, explainable artificial intelligence (XAI) - a set of methods that make the conclusions of AI understandable.
The speaker gave basic architectural principles and end-to-end patterns of the GenAI platform in EDMS: RAG systems and knowledge bases to increase accuracy and reduce hallucinations, multimodality (processing text, sensor readings and images), isolation of sensitive data.
In conclusion, the speaker gave a definition. The evolution of the EDMS is the journey from paper folder chaos to managed documents as a value in its own right. Modern EDMS is a metasystem that turns information into a real result using AI agents.
During the break and at the end of the conference, the participants talked informally, and also had the opportunity to familiarize themselves with the solutions and services of IT suppliers at the stands deployed in the event hall.
| The site content is translated by machine translation software powered by PROMT. The machine-translated articles are not always perfect and may contain errors in vocabulary, syntax or grammar. Read original article If you find inaccuracies or errors in the results of machine translation, please write to editor@tadviser.ru. We will make every effort to correct them as soon as possible. |