AI Assistant for the Road and Motorway Directorate: Tailored Digitalization of Guideline Management
About The Customer
The Road and Motorway Directorate (ŘSD) is a state enterprise established by the Ministry of Transport of the Czech Republic. Its core mission is to manage the construction, modernization, maintenance, and repair of motorways and Class I roads and their related infrastructure, in accordance with Czech road legislation.
In fulfilling its duties, ŘSD works with a large volume of technical regulations, internal and ministerial guidelines, methodological instructions, and other documents defining rules for project preparation and execution, asset management, and public procurement administration.
Challenge
Despite having a document management system in place, employees faced several issues:
- The search functionality relied solely on full-text search, which was insufficient for retrieving relevant information, especially across various document formats.
- Complex legal and technical terminology made it difficult to interpret the content.
- It was essential to ensure access to both current and historical versions of regulations.
This led to delays in locating the right information, a higher risk of misinterpretation, and lower overall efficiency.
Requirements
ŘSD needed a modern and efficient solution that would:
- Enable fast and accurate semantic search within documentation.
- Allow users to ask questions in natural language and receive clear answers.
- Offer transparent verification of answers with references to the original documents.
- Ensure full security and compatibility with the existing IT infrastructure.
What did Cloudfield offer to ŘSD?
By partnering with Cloudfield, ŘSD gained access to a solution combining analytical and technological services with a strong focus on:
- Deep expertise in AI and document processing, fully customized to client needs.
- A flexible delivery approach allowing seamless integration with existing infrastructure.
- Security guarantees, including full control over the system and minimal vendor lock-in.
Implementation
The solution was built on Cloudfield’s internal DOC BRO platform—not as an off-the-shelf product, but as a set of tools and methodologies tailored to ŘSD’s specific requirements.
Key implementation elements included:
- Semantic search enabling users to ask questions in natural language and receive contextual answers.
- Document version control – the AI assistant automatically accounts for the document’s validity at a given date.
- OCR technology – enables processing of scanned and historical documents.
- Integration with Azure Entra – maintaining existing user access rights.
- Hybrid architecture – combining on-premises infrastructure with Azure-based AI models.
- Verifiable answers – users receive direct references to specific guidelines or related documentation, ensuring accuracy and traceability.
Outcome and Benefit
- Faster document workflows – semantically precise responses save employees significant time.
- Minimized errors – verified information with references to original sources improves accuracy.
- Smooth process integration – users continued working in their familiar environment, now enhanced with AI capabilities.
- Security and control – ŘSD maintains full system control without reliance on third-party vendors.