Марія Бровінська AI Eng 15 December 2025, 11:37

What will be the development of AI in 2026? SoftServe experts identify three most realistic directions

Artificial intelligence is becoming the foundation on which businesses build their processes and products. According to PwC, these technologies have the potential to add up to 15% to global GDP over the next decade—one of the largest economic impacts in modern history.

In 2026, AI will continue to evolve, with companies increasingly integrating it into manufacturing, development, and data. Technology experts at SoftServe have identified three areas of AI that will change how companies automate manufacturing, create digital products, and work with information.

Physical AI: synergy of generative AI and hardware

Physical AI is a major leap forward in robotics and autonomous systems. With the help of generative AI, machines learn to perceive, understand, reason, take gravity, friction into account, and act in the physical world as naturally as humans. In 2026, Physical AI will drive transformations across many industries — from autonomous mobile robots that navigate complex environments and avoid obstacles, to manipulators and surgical robots capable of performing precise and complex tasks.

Humanoids and collaborative robots deserve special attention, but not only them - AI will also significantly enhance work with simulations. All autonomous systems can be quickly trained and tested in simulations and digital twins based on synthetically generated data.

“For one of our clients, for example, we created a solution that reduced the simulation time of a production line from several hours to 5 minutes per cycle. In world practice, this approach increases the efficiency, safety, and speed of technology implementation in a real environment: autonomous mobile robots navigate in complex spaces, manipulators adjust their grip to the shape of objects, and humanoid robots become reliable partners of people,” explains Lyubomyr Demkiv, Head of Robotics at SoftServe.

According to Gartner, 80% of warehouses worldwide will use robotics by 2028. At the same time, solutions based on Physical AI will be more accessible — they will be offered by half of the top AI service providers.

Multi-agent systems: a new logic for software development

The amount of data and the complexity of products are growing faster than teams can scale. In such conditions, humanity is taking a step forward - towards multi-agent systems, where not one universal AI performs tasks, but dozens of specialized agents cooperate among themselves, distributing work in the same way as a team of people does.

“What is happening with multi-agent systems now is a transition from AI tools to true AI collaboration. We see how agents can take on entire development stages: requirements generation, code writing, testing, security auditing. This is a change in the work model, where humans focus on complex solutions, and specialized agents perform routine tasks,” says Zoryana Doshna, VP of Advanced Technologies and Gen AI Lab Leader at SoftServe.

The demand for such solutions is growing very quickly: the AI direction at SoftServe is showing 85% annual growth, and over 150 specialists are already working on AI projects, from Data Scientists to Agent Engineering specialists. The created agents analyze technical documentation, propose architectural solutions, write modules, form unit tests, and prepare final technical documentation. Depending on the scenario, this helps to reduce the duration of work by 30 to 70%.

"Our goal is to take software development to a new level using multi-agent systems, so that it becomes a reality, and not just an interesting experiment with AI. In particular, we have created a solution that allows not only to launch individual agents, but also to manage their interaction, track the quality of solutions, and automatically integrate results into DevOps processes. This is the foundation for AI development in the coming years," notes Volodymyr Karpiv, R&D Director of SoftServe.

Multimodal AI: A New Level of Data Understanding

Generative models have become a common tool for business in two years — they write texts, summarize information, and help with communication. However, most real business processes work with completely different types of data — photos and videos, drawings, document scans, tables, and presentations.

That is why the next stage of development is multimodal AI, which can see different data formats and combine them into one coherent context. In SoftServe, this is implemented through a solution created together with NVIDIA - Multimodal RAG. The technology simultaneously analyzes text, images, tables or diagrams and generates an answer based on all data sources. This approach increases the accuracy of results by more than 70% and reduces the time to search for information by approximately 40%. For teams working with large document arrays, this means a significant reduction in manual work and significantly faster decision-making.

In the coming years, multimodal AI will become the foundation for automating processes in finance, industry, medicine, and logistics. It enables businesses to work with data the way experts do: see the big picture, assess context, and make decisions based on all available information, not just one format.

About SoftServe

SoftServe is the largest global IT company with Ukrainian roots, operating in the field of software development and consulting services. Over 10,000 employees work in more than 60 offices located from San Francisco to Dubai.

The company is headquartered in Austin, Texas, USA. The company has over 900 active projects for clients in North America, the EU, and Asia. SoftServe's clients include IBM, Cisco, Panasonic, Cloudera, Henry Schein, Spillman Technologies, and others. SoftServe is a partner with Google Cloud Platform, Amazon Web Services, Microsoft Azure, NVIDIA, and other major technology companies.

The former head of Microsoft Ukraine predicted which IT professions will disappear due to the development of AI and gave valuable advice