Unearthing Lost Knowledge: Scaling Code Understanding With GenAI Through CodeConcise

Legacy systems often resemble archaeological digs, with invaluable knowledge buried deep within complex, aging code. This “lost knowledge” inflates risk, and makes modernisation a daunting prospect. At Thoughtworks, we’re focused on moving beyond the hype by practically applying AI across software delivery. One such initiative has led to the creation of CodeConcise, a tool designed to systematically unearth this hidden wisdom and scale code understanding within legacy systems.

We will explore the journey of CodeConcise, detailing how it has evolved to leverage a potent combination of Large Language Models (LLMs), Retrieval Augmented Generation (RAG), Static Code Analysis, and Abstract Syntax Trees (ASTs). The focus will be on how these technologies work in concert to transform opaque code into accessible insights.

Through this exploration and insights from a real-world case study, attendees will discover:

* The evolutionary path of CodeConcise in its mission to effectively describe and demystify complex code.
* Practical techniques to leverage AI-driven tools like CodeConcise for significantly scaling the recovery and dissemination of critical system knowledge.
* The transformative role such tools can play in navigating legacy modernization, empowering teams to make informed architectural decisions.

This talk is for architects and developers tasked with deciphering legacy systems, and for any technical leader aiming to amplify their team’s ability to understand and evolve critical software assets. Join us to learn how GenAI is becoming an indispensable ally in reclaiming our understanding of the systems we depend on.