Published on November 16, 2023, 11:02 pm

Jay Hack, an AI researcher specializing in natural language processing and computer vision, discovered the potential of large language models (LLMs) like OpenAI’s GPT-4 or ChatGPT to enhance developers’ productivity. With a background as a machine learning engineer at Palantir and having founded an AI-powered shopping startup called Mira, Hack began experimenting with LLMs to automate software engineering tasks such as executing pull requests.

Driven by this idea, Hack and a small team developed Codegen, a platform that leverages LLMs to automate mundane and repetitive software engineering tasks, allowing companies to focus more on product innovation. Codegen goes beyond code-generating AI tools like GitHub Copilot or Amazon CodeWhisperer by tackling codebase-wide challenges such as large migrations and refactoring.

The key differentiator for Codegen is its multi-agent system for complex code generation. This system enables collaboration among agents that collectively solve larger tasks. The use of multiple LLMs working together results in significantly better outputs compared to solo models.


Comments are closed.