Unlock the Potential of Gen AI in Modernizing Legacy Tech

Improving Business Outcomes

Companies have often used gen AI in a simplistic way by directly translating legacy code into modern language. This "code and load" approach simply shifts the tech debt to a new context. The true goal should be to improve systems and processes to generate more value. Gen AI can help understand existing code, determine business needs, and modernize necessary processes. For instance, at one financial-services company, experts' transcripts were fed into the gen AI model to provide better guidance, enabling business and engineering experts to work together.

By using gen AI, engineers can gain clarity on system operations and allow business experts to make more informed decisions. This leads to a more efficient use of resources and a better alignment of technology with business goals.

Enabling Autonomous Gen AI Agents

In software development, using gen AI agents to assist developers can increase productivity. The next level of acceleration involves deploying hundreds of autonomous gen AI agents with human oversight. These agents have distinct roles and expertise, collaborating on complex tasks such as data analysis, integration, testing, and outcome refinement.

For example, at a banking company, the deployment of gen AI agents allowed for a 40 percent reduction in the estimated time to migrate mainframe components. Data mapping and storage agents worked with other specialized agents to ensure safe and secure code development. Implementing controls like feedback loops and ID assignment helps ensure the agents deliver the right outcomes.

Focusing on Scaling Value

The excitement around gen AI has led companies to focus on tool evaluation. However, the bigger issue is how to scale gen AI. Technology leadership should develop a central, autonomous gen AI capability with two components - a factory and a platform.

A factory is a group of people who develop and manage gen AI agents for specific end-to-end processes. It standardizes and simplifies agent development and management. A platform provides reusable services and capabilities that factories can access. It includes a user interface, APIs, supporting services, and a library of gen AI agents.

By focusing on scaling value, companies can build sophisticated multiagent workflows and drive innovation and value.

Next Steps

Companies looking to adopt the multiagent orchestration model should take four steps. First, question any technology proposal with a long timeline and many people. Second, focus gen AI on the biggest problems. Third, tie the business plan explicitly to value and track it closely. Fourth, get ahead of talent, technology, and operating-model implications.

Companies have only begun to scratch the surface in applying gen AI to modernize legacy technology. By focusing on orchestrating gen AI agents on meaningful business opportunities, they can cut tech debt and drive innovation.