Modular AI Agents and the New Operating System of Finance

Today’s modular, API-driven AI agents have the potential to set a new standard for operational excellence, risk management, and client engagement. These systems use a flexible, modular LEGO-style of architecture to enable capabilities such as scalable data connectivity and advanced explainability, for example.

In this article, we’ll explore why modular infrastructure is a solid foundation for AI systems, and as an illustrative example, will highlight a set of specialized FactSet modules that integrate seamlessly with workflows and simplify the creation of AI systems.

APIs and MCP as the Groundwork for Agentic Workflows

Agentic AI depends on connected systems, and APIs (Application Programming Interfaces) are essential. APIs serve as the backbone for connecting Large Language Models, generative AI, and data management systems like data warehouses and data lakes and are crucial for businesses aiming to leverage AI without overhauling existing infrastructure.

That’s why firms are seeking partners who offer flexible, API-first platforms designed to integrate, not isolate, within existing workflows. Notably, over 80% of internet traffic is flowing through API channels, up from just 40% a decade ago. That growth is forecast to multiply rapidly as new AI technologies are built for greater speed, learning capacity, and scale.

As a result, the modular infrastructure choices firms are making right now will eventually separate industry leaders from laggards. The payoff includes the potential for faster innovation, the ability to keep up with regulatory requirements, and a foundation for experimentation and iteration. 

Source: SALT Security, July 2025

Enhancing the APIs that provide the foundation for connecting systems, Model Context Protocol (MCP) standardizes how AI agents access, share, and apply data in real-world workflows. Given agents must reason, plan, and act across multiple data environments, MCP ensures context and coordination.

By embedding context into every interaction, MCP enables agents to maintain continuity across API calls so that decisions are informed by the full market, regulatory, or client background. Embedded context also reduces integration friction by providing a common language for data access, making it easier to expand workflows with new modules without custom engineering. And because entitlements, licensing, and traceability are enforced at the connectivity layer, firms gain auditability and compliance by design rather than a future add-on.

MCP is also important for multiple specialized agents to collaborate such as when handling risk, compliance, or portfolio analytics, for example. MCP enables them to share information securely and consistently, transforming isolated point solutions into orchestrated workflows.

Overall, if APIs are the nervous system of agentic AI, MCP is the circulatory system that makes the APIs interoperable and ensures reliable, governed context flows across every module and workflow. For financial services, MCP means agents can seamlessly weave together proprietary data sources, regulatory checks, and market intelligence feeds without sacrificing control. It allows organizations to treat every system, whether legacy or cloud-native, as a composable building block that agents can dynamically call upon.

For example, the MCP connectivity layer provides:

• Regulated, audit-ready access to financial data

• Secure enforcement of entitlements and licensing

• Flexibility across hybrid cloud, on-premises, and distributed deployments

Natural Language Interfaces

Traditional financial research has historically relied on static reports, which can later make it difficult to address follow-up questions or revisit new data. Given today’s fast-paced environment where financial professionals regularly need updated information and content, a more interactive and continuous approach is increasingly important.

Conversational APIs are a step toward meeting those needs. They enable users to use natural language (similar to how people have conversations) to request information, ask questions, and pull context directly from data platforms. By enabling natural language interfaces that connect directly to a firm’s data and analytics platforms, solutions like the Conversational API powered by FactSet Mercury (our GenAI LLM) provide advanced Q&A chats and contextual insights that are integrated into the tools analysts are already using. Below is an example screen view.

Explainability

Portfolio managers and compliance officers require clarity on where financial data comes from, with assurance that outputs can be audited through traceable, verified sources.

Leading agent systems now leverage explanation APIs that automatically show, for example:

• The underlying drivers of a company’s performance over time (like earnings revisions, M&A, and regulatory changes)

• How company performance relates to peers, sectors, and the macroeconomic environment

• The data that was referenced, including links to the underlying sources for trust and transparency

The following screen image from the FactSet Security Explanation API illustrates an agent describing security underperformance. This advanced explainability turns AI from an enigmatic assistant into a trusted colleague.

Integrated, Real-Time Intelligence

A truly modern agent doesn’t just provide answers; it delivers actionable intelligence into the workflows and dashboards that analysts, traders, and compliance teams use daily.

For instance, consider a portfolio manager evaluating mutual funds. With AI-powered embedded dashboards, they can view a fund’s NAV, 1-year performance, category, rating, and AUM in a continuously updated panel that reflects real-time market changes. Instead of manually cross-referencing multiple sources (and static reports), the PM instantly sees a concise, actionable summary, such as the example below where a widget has threaded together multiple data points automatically for the user.

Similarly, equity analysts can track a stock against relevant benchmarks such as the FTSE 100 index, as shown in the following screen image. Interactive chart widgets provide contextual insights at a glance, enabling analysts to identify trends, compare relative performance, and make decisions without leaving their workspace. The visualizations are more than charts; they are integrated intelligence layers that connect AI insights directly to the data, research, and compliance information underpinning the analysis.

With this modular, embeddable approach:

• Portfolio dashboards update in real-time with the latest risk signals

• Compliance flags appear as interactive widgets linked to source documents

• Sustainability comparisons and developments are annotated with insights

Enriching Unstructured Data

Earnings transcripts, regulatory filings, news articles, and research reports often exist in unstructured formats that are difficult to analyze at scale. Modular AI agents resolve that by transforming unstructured content into AI-ready data that results in actionable intelligence.

By converting text, tables, and complex documents into structured, machine-readable formats, firms can feed insights directly into models, dashboards, and conversational workflows. This enrichment layer ensures that AI systems operate on comprehensive, verified, and context-aware information, dramatically improving both speed and accuracy of decision making.

AI enrichment + Knowledge Graph = actionable intelligence

Enrichment adds meaning: entities, context, and connections

Company data enriched with sentiment, confidence metrics, and multi-level thematic tagging. From governance and human capital to lawsuits, tariffs, and market shifts.

Where Are Leaders Investing? (And Why Should You Care?)

The next decade in finance and technology won’t be defined by who has the flashiest AI model, but by who builds the most adaptable, reliable operating system of AI agents that seamlessly connects data, workflows, and people. The architects reimagining this system are the ones shaping its future trajectory.

Forward-thinking firms are already moving beyond one-size-fits-all approaches. They are embracing modular systems designed for seamless natural language search, conversational workflows, robust data governance, and transparent, audit-ready insights, solutions precisely tailored to the needs of their teams. As a result, they are positioning themselves to deliver smarter insights, faster decisions, and more resilient client experiences.

For more information on APIs, explore FactSet’s Developer Portal.

This blog post is for informational purposes only. The information contained in this blog post is not legal, tax, or investment advice. FactSet does not endorse or recommend any investments and assumes no liability for any consequence relating directly or indirectly to any action or inaction taken based on the information contained in this article.