GIC

Building an AI platform for investment research

BACKGROUND

GIC is a sovereign wealth fund with over US$800 billion in assets under management and 11 offices across key financial hubs worldwide. Driven by recent advances in AI, it aims to build an innovative system that enables investment professionals to conduct research at scale, gather actionable insights from proprietary datasets, and generate alpha through data-driven strategies.

MY ROLE

As a designer on the AI team, I helped shape the product vision for ChatGIC, the organization's first enterprise AI platform. Through close collaboration with data scientists and engineers, the team developed an intuitive experience that empowers users to accelerate their investment research workflows and operate with greater efficiency through AI integration.

Prompt to begin

ChatGIC represents a new approach to investment research at GIC. The platform introduces a chat-first interface where investment professionals begin their journey through natural language interactions, similar to having a conversation with an expert assistant. In addition, users can perform diverse tasks from stock analysis to due diligence within one platform, eliminating the need for multiple systems and significantly reducing research time.

Integrating sources

Integrating Sources

Investment professionals need access to diverse data sources to make informed decisions. To ensure that the AI platform draws on accurate and audited data sources, I designed a unified data integration layer that seamlessly connects to internal proprietary databases, external market feeds, and third-party research platforms.

The feature also intelligently processes financial data with content such as news, reports, and regulatory filings. This allows analysts to quickly surface relevant insights and signals from multiple sources instead of manually piecing information together from disparate systems.

Introducing voice mode

To enhance productivity during research and analysis, we expanded ChatGIC's capabilities to include voice interactions.

The voice mode integrates advanced natural language processing to accurately transcribe industry specific terminology and financial jargon. It also provides real time summarization of long form content and meeting discussions, allowing analysts to focus on strategic thinking rather than note taking. This feature has been particularly valuable during lengthy board and client meetings, where rapid information collection is crucial.

Building a prompt library

To unify the needs of various departments on a single AI platform, I designed a centralized repository where users can access a curated collection of prompts tailored to various aspects of the investment and deal workflows.

This library is a source of inspiration and guidance, offering prompts that provide an easy starting point to accelerate research tasks while maintaining consistency across teams.

Better version control and tracking

Investment research with AI often involves an exploratory process with multiple prompts and conversation branches. We observed that analysts frequently used a trial and error approach, experimenting with different prompts until finding the most effective path. Users would often save promising responses manually and restart conversations to build upon successful interactions.

To streamline this workflow, we introduced version control concepts borrowed from Git such as commits and branches. Users can now save key milestones within their conversations and create multiple branches from these points. This allows them to either continue productive lines of inquiry, revert to previous states if needed, or spawn new conversations that retain the context of successful prompts.

Improving research and report drafting

One of the key challenges we faced is in developing modes. As the product had to serve a wide range of users across different disciplines, it was necessary to include different modes that cater to their specific needs.

The most commonly used mode is featured above. It is where the majority of investment analysts begin their research and explore investment opportunities. These would later evolve to deeper investment insights and detailed financial models that are produced in the form of a generated report.

Result and impact

80%

Adoption across the organization

56%

Improvement in average time to prepare data visualizations

24%

Reduction in average time spent to complete investment reports

ChatGIC launched in mid 2025 and quickly achieved widespread adoption across GIC's investment teams. As usage grows, the platform continues to evolve by leveraging decades of proprietary investment knowledge and expertise. This data-driven approach enables GIC to harness collective intelligence at scale, strengthening its competitive advantage in the global investment landscape.

Figures here are estimates, detailed metrics can be shared in confidence.

NEXT CASE STUDY

Endowus

Designing a recurring investment feature

Read Now