Dar facilitates access to critical data via PARA Copilot utilizing Azure AI on PARA OS

Dar is a global consultancy firm that helps clients plan, design and manage buildings and infrastructure more effectively. With a portfolio of over 4,500 projects successfully delivered to more than 1,000 clients, Dar operates across the Middle East, Africa, Asia, the United Kingdom, and Europe, employing over 10,200 professionals across 57 offices. Dar is also the founder of the global collaborative Sidara, which brings over 21,500 employees across more than 300 offices and 20 industry-leading brands in architecture, engineering, and construction consultancy. “We handle complex projects such as airports, bridges, hospitals, and universities. We’ve even helped plan cities,” says Faysal Shair, Director of Digital Solutions and Services at Dar. “Our digital solutions specialists strive for innovation with impact, empowering clients by crafting digital solutions that enhance sustainability and liveability within the built environment, driving better outcomes for people and planet.”

Managing complex data across systems

PARA OS – Dar’s cognitive digital twin and building portfolio management platform – was developed to empower organizations to make the most of their data, mainly to find ways to enhance the sustainability, resource efficiency, usability, and adaptability of their built assets. “PARA OS targets the entire data lifecycle, collecting data from different systems before processing and presenting it to the user,” explains Shair. The platform seamlessly integrates with various enterprise systems, IoT devices, and automation and control systems to unify data from multiple sources, providing a holistic overview of a building’s data. 

By offering a unified data platform environment with advanced visualization capabilities, PARA OS provides easy access to centralized asset information with single-source visibility of multiple building and infrastructure assets. The modularity of PARA OS allows for customized solutions that span several data types and insights. These modules include portfolio management, energy intelligence, sustainability, occupant wellbeing and asset management. 

With all modules generating rich data in a single platform, PARA OS offers a powerful, centralized experience, thus making thoughtful design and user guidance essential for maximizing its benefits “The data PARA OS contains is a gold mine,” remarks Nawal Zaher, Analytics Manager, Digital Solutions and Services at Dar. “As we enriched PARA OS with more data, insights, and engineering expertise, the platform grew significantly in value. To ensure users can easily access the right information amid thousands of data points across multiple assets, enhancing discoverability became a key focus.”

"We wanted to enhance the user experience by personalizing interactions, accelerating data delivery, and boosting the discoverability and accessibility of insights,” explains Zaher. “This strategic focus led us to partner with Microsoft and leverage Azure AI Services.”

Simplifying accessibility of critical data

Dar has developed a powerful multi-agent system that integrates with PARA OS, enabling users to interact with the platform in natural language via text or voice. With the Azure AI Services, Dar was able to overcome challenges related to discoverability and information retrieval.  “Azure AI powered PARA Copilot allows users to interact with PARA OS asking questions in natural language, searches through all the documents, stitches the relevant parts into an answer, and delivers it in a fraction of a second.” highlights Zaher.

To improve this functionality, Dar used multiple Microsoft technologies. “We were able to enhance features in PARA Copilot using Azure OpenAI, Azure AI Services—specifically Computer Vision—and Azure AI Foundry for deploying and fine-tuning the AI model,” notes Zaher. These technologies assisted PARA Copilot to process complex queries, extract relevant data, and strengthen accessibility for users managing large-scale projects.

Multi-agent system enhances speed and accuracy

Users can retrieve data and generate graphs of trends all through one interface. PARA Copilot is built on a multi-agent system (MAS) that routes user requests to the appropriate module for faster processing. When a query comes in, it goes through a supervisory agent which directs the query to different specialized agents, as needed. For instance, the SQL agent retrieves data relevant to the query, an information retrieval agent processes documents using retrieval-augmented generation (RAG), while a memory agent tracks conversations within the same session. 

“PARA Copilot responds to queries in 3 seconds or less, enabling users to quickly access key data,” highlights Zaher. “And with Azure Computer Vision, we’ve seen a 25% improvement in the accuracy of processing of handwritten documents, to extract insights from maintenance records, widening our knowledge base.” The engagement change has also improved the accuracy of the SQL query agent by 30%.

PARA Copilot also enables users to ask the platform to look for correlations through simple queries such as "Which factors are negatively affecting occupant wellbeing the most? Give me the relationship between these factors.” Drawing comparisons with historical data has also become simpler — users can simply ask PARA Copilot to compare two values or map trends on a graph over a specified time frame. They can then ask follow-up questions on data correlations. “These correlations and calculations provide powerful insights that users would not be able to extract as easily by navigating PARA OS manually,” Shair explains. “Now, users can access these insights with simple prompts, making complex data analysis more accessible, and therefore, actionable. With PARA Copilot, our engineers specifically noted a 50% improvement in discoverability of information, enabling operators and other stakeholders to have a real-time conversation with their buildings about operational performance.”

Along with speed and accuracy, PARA Copilot improves accessibility and ease of use, reducing the effort required to navigate the application. “The voice activation feature is not only convenient but also improves PARA’s accessibility for people with disabilities, eliminating the need for manual input,” notes Shair. 

Training new users has also become simpler. By enabling users to interact naturally with the system, the PARA Copilot makes it easier for facility managers and operators as well as both technical and non-technical users to retrieve relevant data and make informed decisions.

Dar’s digital services specialists deployed GitHub Copilot to breeze through repetitive coding tasks, write cleaner and more efficient code, and speed up the entire development process. The real-time code suggestions helped developers minimize errors and bugs, boosted productivity, made collaboration among team members smoother, and even cut down on development costs. 

Maximizing potential for the future

Dar continues to refine PARA Copilot. “Our vision entails transforming PARA Copilot into a software-as-a-service offering that can be added to applications beyond PARA OS and made available through Azure Marketplace. In addition to this, our intent is to have PARA Copilot deployable with on-premise and hybrid schemes as well thus supporting extended number of clients,” shares Zaher. “We are also exploring additional features for the CUI, such as reinforcement learning based on human feedback, and enhancing the experience by suggesting more tailored prompts to users.” 

As it looks to the future, Dar remains focused on empowering users with a platform that simplifies complex data and enables more effective decision-making. “The pursuit of continuous improvement is the road to innovation. Engaging with Microsoft, a leader in cutting-edge technology and leveraging Microsoft’s advanced AI Services, has broadened our vision for the CUI,” says Shair. “This collaboration not only accelerates our journey towards creating a more intuitive and impactful user experience but also pushes the boundaries of what’s possible and opens up pathways for next-generation advancements.”

Discover more about DAR on Facebook, Instagram, LinkedIn, and YouTube.