AI Agents Are Changing Construction
At the international BUDEXPO-2026 construction exhibition in Minsk, Vitro-Soft presented the results of AI agents running on its Vitro AIR platform, integrated with the Vitro-CAD common data environment (CDE). The solution automates routine checks that previously required specialists to spend long hours on painstaking manual work.
The construction industry incurs multimillion-dollar losses because data is fragmented and too much work is done manually. Document reviews can drag on for weeks, while errors lead to downtime and penalties. Information is scattered across building information modeling (BIM) systems, common data environments (CDEs) and Excel spreadsheets, so aggregating it can consume up to 30% of working time. However, conventional AI systems are difficult to implement because companies often lack both machine-learning engineers and the computing resources they need.
A practical way forward is offered by AI agents. They automate data analysis without requiring complex IT infrastructure, synchronize information from different sources and identify design errors before construction work begins. This helps projects stay on schedule and reduces the risk of construction defects before they can become costly problems on site.
Vitro AIR: AI Without Coding
Vitro-Soft LLC is a leading Russian developer of project data management solutions for the construction industry. Its flagship product, Vitro-CAD, creates a common data environment for collaborative BIM projects. The platform integrates with industry-specific software, including Revit, nanoCAD and Renga, and uses dual versioning to keep draft documents strictly separate from approved versions.
The development of this technology led to the concept of CDE 2.0. Rather than simply storing files, this approach manages the contextual information behind a project. Artificial intelligence takes over routine data processing, allowing professionals to focus on engineering decisions.
Vitro AIR puts that concept into practice and extends the company’s ecosystem. While Vitro-CAD organizes project data and controls access to it, Vitro AIR adds an intelligence layer. The no-code platform lets users create autonomous AI agents that connect to different language models. Ready-made templates tailored to the construction industry allow a basic assistant for checking documentation to be launched in just two to three days. The architecture also offers deployment flexibility: users can work with cloud services or install models locally on their own servers.
Automated BIM Model Checks
Tests conducted between November 2025 and June 2026 showed substantial reductions in the time required for routine checks. Processing a package of as-built documentation in PDF format takes one minute, compared with 40 minutes previously. A check of BIM models in IFC format, covering 42 validation rules across 19,442 objects, takes eight minutes. The platform also supports the Information Delivery Specification (IDS) format, comparing a model’s actual parameters against client requirements expressed in a machine-readable form.
Although the claimed 40-fold acceleration remains based on internal testing, Vitro AIR has already received top recognition from industry experts. The platform took first place in the “AI in Construction” category at the TIM-LIDERY (BIM Leaders) 2024/25 competition.
Exporting Russian AI
Vitro-Soft’s new solution has export potential backed by practical experience. Vitro-CAD has been deployed more than 100 times in Belarus and Kazakhstan, including at Minskmetrostroy’s design institute, Minskmetropolitanproject. In August 2026, a pilot AI module for checking documentation completeness was tested in Kazakhstan, cutting the process from five days to two. Support for the open IFC and IDS formats also makes integration with international modeling systems easier.
From Design Clashes to KS-2 Forms
Other Russian companies are developing similar solutions.
In 2021, ASCON JSC introduced geometric clash detection for models in its Pilot-BIM platform. In 2022, organizations within Rosatom State Corporation developed a prototype AI system that extracted requirements from documents with 90% accuracy. From 2023 to 2024, Glavgosexpertiza of Russia, the Main State Expertise authority, introduced predictive analytics modules. In summer 2025, the Moscow Artificial Intelligence Center launched its Tsifrovoy Normokontrol (Digital Standards Compliance Check) service in nine regions. In 2026, property developer APRI Flash JSC and Napoleon IT LLC implemented an AI agent to cross-check KS-2 construction work acceptance certificates, reducing the process to 10 minutes. Signal LLC tested algorithms for detecting errors: its model identified 13 of 14 planted errors, but an engineer confirmed only some of the findings, underscoring the need for human oversight.
From Experiments to Commercial Demand
Automating construction documentation workflows is emerging as a distinct segment of Russia’s construction IT market. The construction digitization market was estimated at RUB 26 billion a year in 2025–2026 (about US$310 million). According to forecasts from Russia’s Ministry of Construction, Housing and Utilities, it could grow to RUB 130 billion (about US$1.6 billion) by 2028.
Solutions that use AI to process design and as-built documentation are seeing sustained commercial demand. AI agents on the Vitro AIR platform demonstrate that automating routine checks is no longer just an experiment. It has become a practical tool for design firms, with demand in Russia and international markets alike.
