From 10 Minutes to 30 Seconds, how CSOP is redefining asset management with Microsoft Azure AI

Cenny Cui Haofan’s mornings used to kick off with him opening over 50 Excel files, digging through a maze of emails, and chasing down colleagues across departments, just to compile one Exchange-Traded Fund (ETF) report for the Hong Kong Exchange. It was a daily grind that consumed the CSOP associate’s time and energy, leaving him little room for strategic thinking.

Today, that same task takes just 30 seconds.

CSOP, a leading Hong Kong-based asset management firm known for its ETFs, has embraced AI not just as a tool, but as a cultural shift. With Microsoft Azure AI and GitHub Copilot, CSOP has built an “Intelligence Hub” (the Hub) that empowers every team to create custom AI solutions, no coding required. The results? A 30x leap in efficiency, 99% automation in key workflows, and a workforce now focused on innovation instead of administration.

The Intelligence Hub

At the heart of CSOP’s transformation is the Hub — a dynamic AI ecosystem built on Microsoft’s Azure AI and GitHub Copilot. With over 1,800 pre-built models on Azure AI Foundry and GitHub Copilot’s AI-powered coding capabilities, the Hub empowers every department to build custom AI applications—no coding background required.

Azure AI’s multi-model capabilities handle complex data with precision, while its enterprise-grade platform ensures security, scalability, and compliance—critical in the highly regulated finance sector. GitHub Copilot translates ideas into code, enabling rapid prototyping.

Before, launching a new product meant waiting three months for IT to build the software architecture,” says Melody He, CSOP’s Chief Business Officer and Deputy CEO. “Now with Azure Al and GitHub Copilot, we can prototype and deploy in just a few weeks. That speed lets us respond to market shifts instantly, and our clients feel the difference.

This “try first” mindset is backed by Microsoft’s secure and scalable enterprise-grade AI platform, allowing CSOP to innovate confidently while complying with the market’s strict regulations and governance standards. Now the company runs with improved efficiency, optimized resources, and stronger client relationships.

Trading reinvented

Trade confirmations once arrived in countless formats – some embedded directly in email bodies, others hidden in attachments like PDFs and Excel sheets. This inconsistency requires manual interpretation and long hours. The process was time-consuming, error-prone, and often required traders to work late-nights or early-mornings to keep pace with global markets and international time zone differences.

Emma Wang, Deputy Head of Trading, and her team used GitHub Copilot to swiftly prototype an AI-powered application to extract trade data from documents. They further refined the system by employing multiple large language models on Azure AI Foundry, such as OpenAI’s o1 and o3 large language models as well as DeepSeek’s R1. The AI application now understands the nuances of trade confirmations––no matter the format––and extracts key details like prices, quantities, and settlement dates with remarkable accuracy.

Before AI automation, trade confirmations could take at least 60 minutes on a calm day, stretching into prolonged hours during volatile trading days, with teams manually reviewing hundreds of trades,” Emma recalls. “Now, the AI understands those nuances—saving us significant time and reducing the risk of errors.

With tasks that once took at least 60 minutes now completing almost instantaneously — a nearly 99% reduction in time — traders are now free to focus on what matters most: strategic decision-making and building strong client relationships. The automation allows her team to dedicate their expertise to high-value tasks like gatekeeping, activating strategic operations and setting parameters. This shift enables them to provide quicker, more informed responses to clients, and to efficiently onboard new products, enhancing overall client service.

From chaos to clarity

For CSOP’s Associate Cenny Cui Haofan, the time saved for generating ETF reports is now spent on cross-department collaboration and strategic planning. “My mornings are no longer chaotic. The reports are done in under a minute and I can turn my attention to higher-level tasks.”

Using Azure AI and GitHub Copilot, his team built a user-friendly portal that automated data submission and democratized access to real-time ETF insights. A 30x increase in efficiency has resulted – daily tasks that once took 10 minutes now take just 30 seconds.

Turning data into decisions

With over 50 ETF products across Hong Kong and Singapore, CSOP’s investment team was spending hours each day manually gathering data and drafting reports for frontline sales. Yi Wang, Deputy CEO and Chief Investment Officer at CSOP, developed another AI analytic tool leveraging the Intelligence Hub’s underlying Azure AI and GitHub Copilot capabilities.

The tool analyzes large volumes of stock charts and sell-side research reports to identify market trends and generate trading ideas for internal reference. Monthly reporting effort has dropped by 75%, freeing time for deeper analysis and faster decision-making. Built in days, the solution fits seamlessly into workflows and supports both trading and marketing teams.

We used to spend 1-2 hours a day just on manual extraction,” says Yi. “Now, that time is reclaimed for deeper analysis and faster decision-making, saving 3 to 4 times the effort previously spent on monthly reporting.

Democratizing innovation

At CSOP, innovation isn’t limited to IT, it’s everyone’s business. Through the AI Academy, one-third of the company joined in just a month, learning to build AI apps using “vibe-coding” (natural language programming) on Azure AI. With IT’s support, departments began solving real business problems on their own.  As Cenny Cui Haofan from the operation team puts it, “Now we can basically create an application in just one or two days”, underscoring how accessible development has become.

Every solution is reviewed by the IT team before deployment to ensure security, compliance, and alignment with core AI principles – powered by Microsoft’s trusted AI platform. Thus, a culture where experimentation thrives, innovation is democratized, and responsibility is built in by design.

Agentic AI and beyond

Looking ahead, CSOP isn’t slowing down. With eyes on agentic AI and next-gen investment tools, the firm is building an intelligent, adaptive organization ready for whatever the market brings next. They are particularly interested in further leveraging AI to facilitate investment decisions and better track ETFs and stocks.

Click here to load media