"Microsoft’s deep technical collaboration, coupled with the secure and compliant foundation of Azure—including FedRAMP authorization, strict data residency controls, and geo-restriction capabilities—gave us the confidence to develop within stringent government security frameworks. This partnership with the Microsoft AI Co-Innovation Labs accelerated our ability to turn a complex, high-impact vision into a working prototype ready for real-world adoption." – Ayush Jain, Co-founder and CTO, AssetIntel
...

Co-Innovation Challenge

AssetIntel is an asset management provider in the transportation infrastructure industry, offering solutions to both commercial and government agencies. One of their primary focuses is bridge management, where they help agencies save millions with proactive planning and inspection solutions.

Bridge management demands frequent thorough inspections to help engineers determine bridge maintenance needs. AssetIntel saw the need to solve one major pain point for bridge inspections: manual workflows. Agencies were bogged down with processing large defect and repair datasets, which carried a significant risk of human error. Manual inspections and assessments also caused project slowdowns and increased costs for agency customers.

Seeing an opportunity to improve their bridge inspection solution with AI, AssetIntel decided to partner with the Microsoft AI Co-Innovation Lab to accelerate development, leverage the Lab team’s deep AI expertise, and take advantage of our close alignment with Microsoft’s ecosystem.

In the Lab

The AssetIntel and Microsoft AI Co-Innovation Lab team met in our San Francisco Lab to create an AI bridge inspection solution. This AI-powered system could process and index bridge inspection images, analyze them, and then categorize and generate content for bridge assessments.

Our Lab team chose Azure Computer Vision, Azure OpenAI GPT-4o, and Azure AI Search to power the solution, which achieved the following capabilities:

  • Retrieval of images based on user queries or user uploads
  • Analysis of images against vast image and metadata records
  • AI-generated defect codes and descriptions of potential defects for engineers to review.
arch
Bridge inspection solution architecture

In the lab, the automated image analysis both reduced processing time by over 40% and increased accuracy of defect identification and defect code suggestions. By dynamically integrating GPT-4o for both the analysis and response generation, the AssetIntel team can significantly improve inspection accuracy, reduce reliance on human subjectivity, and streamline the process.

Our collaboration with AssetIntel not only resulted in a working prototype. We also ensured seamless integration in a secure Azure environment, faster deployment, and hands-on enablement of their technical teams through knowledge transfer and training.

Solution Impact

The engagement ended with a successful prototype that significantly enhanced inspection efficiency and accuracy by automating image analysis and defect detection, leading to a measurable reduction in manual review time. This AI tool can help engineers assess and provide accurate and timely bridge maintenance to government and commercial infrastructure groups.

impact

Beyond accuracy and efficiency, the solution being built in Azure’s secure environment helps AssetIntel to meet FedRAMP compliance, ensure data residency, and integrate with geo-restriction controls.

As AssetIntel readies their solution for production, they have joined the FedRAMP marketplace, expanding the reach of this and other solutions to federal agencies. They will also be one of very few companies working with the US Army Corps of Engineers, positioning them as innovators in bridge maintenance.

With a soon-to-be-announced launch of their AI-powered bridge inspection solution, the Microsoft AI Co-Innovation Labs is proud to be a partner in empowering AssetIntel to do more with AI in transportation infrastructure.

立即加入

微软人工智能和物联网实验室

即刻申请，获得与技术专家共同设计解决方案的机会。