How Devessence helped a real estate finance firm streamline mortgage operations with AI. The result: faster processing, improved system stability, and smarter decision-making across teams.
Product
Commercial Mortgage Servicing System
Vertical
Financial professional services
Client
Commercial real estate finance firm
Buyer persona
Company's internal users: underwriters, AML officers, etc
GEO
North America
Provided services
Application development, architecture redesign, performance optimization, quality assurance
Technology
ASP.NET Core, Blazor, Azure Cosmos DB with SQL driver, MS SQL DB, Azure Functions, Azure Service Bus, Azure Blob Storage, Extend, Microsoft Entra, Application Insights, Azure DevOps
Background
As transaction volumes grew, the system needed some enhancements to keep up and avoid delays and inefficiencies. Additionally, there was no streamlined way to distribute fees among brokers and analysts, making negotiations time-consuming.
Our Task
Our team worked on two parallel tracks: enhancing the existing platform used for managing commercial mortgage servicing and developing a new CRM system designed for the client’s sales force.
For the Commercial Mortgage Servicing System, our main focus was on improving stability, performance, and scalability, ensuring the system could better handle daily operations for employees. This platform stored all deal-related data and transactions, and it required upgrades to become more reliable, efficient, and user-friendly, including improved tools for tracking progress, generating reports, and exporting data.
In parallel, we began building the new CRM system. This tool will support brokers and analysts in managing fees, providing transparent insights into deal progress, and facilitating fair fee distribution among all involved parties. It’s designed to improve collaboration and decision-making across teams.
Challenges
During the development, we had to overcome several key obstacles.
Need for documentation
The existing system, built with Blazor and Syncfusion components, had been running in production without proper documentation, even though the underlying technology itself is relatively modern. This made onboarding new developers and maintaining the platform challenging. Understanding dependencies and workflows required extensive reverse engineering..
Scalability and performance limitations
The architecture needed some improvements to improve scalability and maintainability. As transaction volumes increased, performance bottlenecks became more apparent, impacting efficiency.
Integration of new business-specific features
Adding new transaction types and business-specific functionalities required careful implementation to avoid disrupting existing workflows while ensuring compliance with financial regulations.
AI-driven document processing challenges
The development of a new AI-powered document processing system required integrating a complex workflow for analyzing and processing tax documents. The system needed to handle large volumes of data efficiently while ensuring real-time responsiveness and accuracy.
Solution
To automate the ingestion of financial statements related to commercial mortgage servicing and governance, we needed both to refine the existing platform and develop new features. Our approach focused on making the system more reliable, automating key processes, and enhancing user experience.
Refactoring
The system had been running for nearly ten years, and over time, its performance had slowed. To improve stability and speed, we reviewed the entire codebase and identified weak spots. We restructured the architecture to make the system more efficient and easier to maintain.
A major improvement was automating the deployment process. We built a Pulumi application to manage CI/CD, reducing the need for manual updates and minimizing errors.
We also added new transaction types and features specific to the business. These updates allowed employees to process deals more accurately and efficiently. Finally, we improved the system’s overall performance, making it faster and more responsive for daily use.
Developing a new AI-powered document processing system
Processing tax documents manually was slow and prone to mistakes. To solve this, we designed a system that automates document processing using AI. We built an event-driven workflow with Azure Durable Functions, ensuring documents move through the system efficiently.
We integrated third-party AI tools to analyze tax documents and extract useful information. Employees can now track the status of each document in real time through a clear, easy-to-use interface.
To give users better insights, we added charts and statistics. These help employees see deal progress, track fee distribution, and analyze financial data at a glance. The result is a faster, more accurate system that reduces manual work and improves decision-making.
Results
Improved system stability and performance
We fixed critical bugs and deployed new features. It made the system more stable and reliable. Users now experience smoother workflows with fewer disruptions.
Automated deployment for faster updates
We built a fully automated CI/CD pipeline using Pulumi. This reduced the need for manual updates, minimized errors, and made deployments faster and more efficient.
Faster and more accurate document processing
The new event-driven system, powered by Azure Durable Functions and third-party AI tools, speeds up document processing. Users can now track document progress in real time, reducing delays and improving accuracy.
Clearer deal tracking and fee management
The updated UI and workflow enhancements make it easier for brokers and analysts to manage deals. The system now provides a clear view of deal progress, fee distribution, and approvals, helping all parties stay informed.
Better insights for smarter decisions
We integrated detailed charts and analytics, giving stakeholders access to valuable financial data. This allows them to track business performance, analyze deal trends, and make well-informed decisions.
Tech Stack
ASP.NET CORE
Blazor
Azure Cosmos DB (SQL driver)
MySQL
Azure Functions, Azure Service Bus, Azure Blob Storage
Microsoft Entra
Application Insights
Azure DevOps
Palumi
Extend