AI is changing ERP. The impact of AI and automation on the development and personalization of ERP systems [Interview]
AI technology is finding increasingly widespread applications. It's no surprise, then, that ERP system manufacturers are also turning to it to improve their software. Artificial intelligence not only provides support to users of these solutions but also allows for broader use of their functionalities. So, what is the direction of ERP system development under the influence of AI technology?
Nowadays, almost every manufacturer has enriched their software with artificial intelligence algorithms. They include virtual agents, as well as tools based on ML, LLM, NLP, and DL. All this is aimed at making operations and management more efficient, and above all, to the benefit of business. Daniel Duda, Director of Business Solutions at OPTeam, discusses the use of AI in ERP systems.
1. How do you assess the current state of AI integration with enterprise systems in Poland?
While 2024 was a year of intense experimentation in the field of artificial intelligence, 2025 is already the time for its use in enterprise-class solutions. Business software vendors are already making extensive use of AI to improve their offerings, reaching for machine learning and natural language processing. For example, Comarch has long been developing AI and machine learning functionalities, available as extensions to its systems. These include intelligent OCR, accounting and sales assistants, and predictive reporting. Thus, artificial intelligence is becoming a crucial element of ERP systems, significantly increasing their potential for process automation and personalization.
This is clearly demonstrated by several examples where the benefits of AI implementation were first recognized. The first is the optimization of processes, which can be improved based on historical data and the predictive capabilities of AI. The second is the elimination of bottlenecks, which AI can identify more quickly, simultaneously pinpointing their causes and ways to address them. The third is forecasting, which in such a dynamic reality allows companies to better plan purchases and sales and prepare for potential events. The fourth is personalizing customer relationships and communication with them using personalized information and offers. The fifth is supporting team competency building, where HR can utilize AI as a tool to help design employee development paths. It all comes down to analytics, which, with artificial intelligence, proves faster and more accurate. With access to vast amounts of information, AI can process data faster and deliver precise results.
2. Z jakimi oczekiwaniami klientów spotykasz się najczęściej odnośnie AI?
Oczekiwania wynikają przede wszystkim z potrzeby usprawnień, zatem dotyczą możliwych ulepszeń, które mają ułatwiać prace. Zmieniają się w zależności od tego, czy AI ma obejmować całą firmę, czy tylko jej obszary bądź wybrane procesy. Najczęściej koncentrują się na automatycznej analizie dużych ilości danych i automatyzacji powtarzalnych zadań, czyli zdawałoby się zagadnieniach, w przypadku których ewentualne korzyści z wdrożenia AI są najbardziej namacalne.
Dotyczą też samego procesu implementacji, mianowicie, aby AI współpracowała z istniejącym systemem ERP bez jego przebudowy, a w oparciu o integrację przez API lub gotowe moduły. Znacznie częściej firmy chcą zaczynać od małych projektów z czasem rozszerzanych na kolejne obszary funkcjonalne systemu ERP. Skalowalność niewątpliwie ma związek z weryfikowaniem skuteczności sztucznej inteligencji i optymalnym rozłożeniem wdrożenia. Firmom zależy, aby implementacja nie zakłócała bieżących procesów.
3. What are the biggest technical challenges when integrating and/or implementing AI with systems such as Comarch ERP or TimeLine?
Both systems are ready for integration with AI, and the technical challenges will largely be the same. However, it's important to note that implementations using these systems vary depending on the company's industry and the area served by the software. Comarch ERP can encompass all departments of the company, while TimeLine is treated as a domain-specific production solution that works with another ERP system. However, for each area—HR, accounting, sales, production, or otherwise—it's necessary to create appropriate scenarios, identifying key issues. Therefore, it's crucial to determine which processes AI will assist, and then what data we have available to base AI on. Here, the complexity of the processes, often highly unique, and the quality of the data are particularly important, as they must be up-to-date, complete, and standardized, which can be challenging given the dispersed sources. Another crucial issue when integrating AI tools with ERP systems is developing compatible interfaces to ensure smooth interoperability. Furthermore, compliance with industry regulations must be maintained.
4. How do you address security and compliance issues, especially in the context of sensitive data?
Security and regulatory compliance are key when integrating AI with an ERP system. Implementation relies on a range of data, including sensitive data, which is subject to numerous regulations, such as GDPR, SOX, and industry-specific regulations. Therefore, the implementation process includes incorporating security-enhancing technologies such as data encryption, RBAC permission management, an MFA authentication layer, and SIEM to monitor unusual behavior on the AI side. Security testing is also crucial to verify protection against potential attacks. In addition to technology, organizational and procedural measures must be taken, which should involve the data protection officer.
4. What does a typical AI implementation process in an ERP system look like?
The process consists of several stages and begins with an analysis of business needs in the areas where AI is to be implemented. It's a good idea to establish KPIs so that the effects can be measured and potential improvements can be made after implementation. These are very specific, calculable metrics that will precisely show, for example, the percentage of process improvement, the time taken to complete an action, or the level of cost reduction. The next step is process mapping and a quantitative and qualitative assessment of the data available in the ERP system and other interoperable domain solutions. It's often necessary to create a unified, standardized, and secure database for the implemented AI models. Next, the AI solution is designed based on the selected AI/ML/LLM model, technology, and tool. An initial step might involve creating a prototype and sampling its capabilities on a small database, gradually expanding it and assessing the accuracy of the implemented AI solutions. After testing, a decision can be made as to whether the tool is suitable for scaling or requires further modifications. Once ready, integration with the ERP system occurs via a dedicated module or API connection. Production deployment is always preceded by performance, quality, compliance, and security testing, which ultimately determines the launch. It's important to remember that the implementation process for AI tools doesn't end with deployment, as they require maintenance and development just like any other software. Perhaps on a slightly different scale, monitoring and updating AI models are still essential for achieving high-quality results.
5. What specific business benefits do you see for your customers after implementing AI?
The greatest benefit is better and more comprehensive use of existing data, which forms the basis for precise reports. This, in turn, translates into decision-making across all dimensions: operational, financial, and business, allowing for anticipation and easier adaptation to market changes. Another benefit is the automation of repetitive processes, which leads to time savings where previously manual actions were required. As a result, a certain amount of cost reduction is also noted. An ERP system supported by artificial intelligence promotes improved service quality, primarily through workflow automation and communication personalization. I also think it's worth emphasizing faster adaptation to market changes, as AI accurately detects and predicts trends.
6. What advice would you give to companies that are just thinking about integrating AI with ERP systems?
I think this would be less advice than a statement. AI will be similar to ERP systems: first an alternative, then a choice, and finally a necessity. Technology exists to be used. Many companies have already recognized the benefits of combining ERP and AI, leveraging the potential of these tools. Statistics show that this is the right decision. However, it's important to approach implementation consciously, recognize and understand business needs, and prepare technologically and organizationally. Base the AI transformation on a well-thought-out and planned strategy.

Daniel Duda, Director of IT Solutions for Business, OPTeam
He has extensive experience in digitizing management processes. He helps improve productivity and adapt to changing conditions by advising on solutions that best meet identified needs. He has completed numerous projects using ERP, MRP, MES, APS, BI, WMS, and low-code tools.