MongoDB Inc. is expanding beyond its database roots and taking aim at one of the thorniest challenges facing many enterprise IT organizations: outdated legacy applications.

At its MongoDB.local New York City event today, the company is announcing the MongoDB Application Modernization Platform, an artificial intelligence-powered approach to transforming decades-old applications into modern, scalable services. The platform combines tooling, structured modernization methodologies and human expertise to tackle technical debt, a $1.52 trillion problem, the Consortium for Information & Software Quality estimated in 2022. MongoDB said its AMP practice can reduce modernization timelines by up to two-thirds.

“Legacy applications are major blockers that are slowing innovation,” said Shilpa Kolhar, senior vice president of application modernization at MongoDB. “Making the simplest change is extremely risky. The developers familiar with those old applications and the older stack are gone.”

The company said AMP is the culmination of over two years of development and customer tests. The platform addresses not only the database layer but also full application stacks, which often include outdated frameworks, brittle runtime environments and poorly documented stored procedures.

“AMP covers the entire life cycle of modernization,” Kolhar said. “It combines AI-powered tools, methodologies and modernization expertise to help companies rapidly transform their legacy application into modern, scalable services on MongoDB and 10 times faster than any traditional approach.”

MongoDB said Switzerland’s Bank Lombard Odier & Co Ltd., a 200-year-old private bank, cut migration testing times from three days to three hours. Australia’s Bendigo and Adelaide Bank Ltd. reported a 90% reduction in development time needed to move a core banking application from a relational database to MongoDB Atlas.

Not a black box

Kolhar said AMP isn’t a “black box” that feeds old code into a large language model. “That doesn’t work,” she said. “Real applications have large, complex code bases that AI can’t directly handle correctly and efficiently. Our tooling enables us to break this problem into smaller chunks using an interactive and automated process.”

The database has become “the agent’s memory and the source of truth” in intelligent systems, said Ben Cefalo, MongoDB’s senior vice president and head of core products and Atlas Foundational Services. “The database is the agent’s memory and the source of truth, allowing it to operate coherently across multiple cycles and even coordinate with other agents,” he said.

Cefalo said MongoDB’s JavaScript Notation-based document model, integrated with vector and text search, is well-suited to AI because it aligns with how large language models and AI frameworks consume and emit data. “If you were to ask AI what the ideal database looks like in this agentic world, you would find that it looks a lot like MongoDB,” he said.

Photo: Robert Hof/SiliconANGLE