At the core of virtually every large organization lies an aging digital infrastructure, often slowing down innovation and agility. In fact, 70 percent of the software powering Fortune 500 companies was developed 20 years ago. Or more. Yet modernizing these critical systems can be incredibly complex, requiring hundreds of engineers, multiyear timelines, and millions of dollars, depending on the amount of code—all while risking obsolescence before completion.
Additionally, the business processes intertwined with these legacy systems further complicate modernization efforts. They have been tailored over decades to fit the existing infrastructure, and often require significant re-engineering and optimization to align with new technologies, adding another layer of complexity and risk to any transformation.
QuantumBlack, AI by McKinsey
Our team from QuantumBlack, AI by McKinsey, set out to change this equation with LegacyX. Our proprietary platform uses the power of generative and agentic AI to accelerate modernization efforts by up to 80 percent while reducing costs and resource intensity. It’s increasingly being used to refresh current applications as well.
These autonomous AI agents not only enhance the technological framework but also re-engineer and optimize underlying business processes. By intelligently analyzing and adapting workflows, agentic AI ensures that business processes are seamlessly integrated with the modern infrastructure, thereby driving efficiency, reducing operational bottlenecks, and enabling more agile and responsive organizational operations.
From generative AI to agentic AI: A game changer for legacy systems
When generative AI became ubiquitous in 2023, one of its earliest use cases was updating software code. “With the unprecedented power of large language models, trained on software development, suddenly it became feasible and economically viable for organizations to attempt these kinds of modernizations that previously had been too difficult and expensive,” explains Dan Collins, associate partner and global asset leader for LegacyX. Early on in projects, they saw a 20 to 30 percent improvement in developer productivity and acceleration.
But agentic AI—the next evolution—has taken modernization to a whole new level. Instead of handling single tasks, agentic AI orchestrates autonomous squads of specialized agents performing different roles that can assess, update, and build complex technology infrastructures in real time.
“The progress we’re making with gen AI is not just a hop, skip, jump,” observes Dan. “We’re advancing thousands of miles with every leap and iteration of the technology on a scale we’ve never seen before.”
In one recent example, McKinsey partnered with a multinational bank that was in the middle of manually converting a fragmented landscape of more than 100 legacy risk models from SAS to Python.
The team deployed the LegacyX platform into the client environment and developed five squads—with 25 AI agents each—performing different roles, such as data engineer, coder, and cloud expert. Here’s how the project unfolded:
- The agents reviewed the legacy code and translated the business logic into plain English documents
- Business and IT experts evaluated the existing state based on the documents and agreed a revised “target” state.
- Agents then converted the models into Python with 90 percent accuracy
The result? An 80 percent acceleration in project timelines while keeping human developers in the loop.
Beyond code: Unlocking AI for everyone
McKinsey has developed similar use cases across industries—including banking, healthcare, and automotive.
The guiding ambition is to enable all talent to use the platform’s capabilities. In an early experiment, business analysts with no technical experience were able to create squads of agents to synthesize interviews and test hypotheses, proving that agentic AI isn’t just for developers—it’s a productivity multiplier across entire organizations.
“The real breakthrough is accessibility,” says Dante Gabrielli, principal product manager. “A user can define objectives, structure the ideal team of temporary experts, and deploy them to execute seamlessly—without increasing the need for additional human resources.”
McKinsey’s Never Just Tech approach is about more than deploying AI—it’s about creating human-led, AI-powered solutions that drive meaningful business transformation. LegacyX is a testament to this philosophy, helping organizations reinvent, rather than replace, their legacy systems—with intelligence, speed, and scale.