As we engage with the broader tech community, it's crucial to clearly articulate what we're building at NonBioS, how our system works, and how users interact with it.
Consider the typical workflow of a software engineer in a conventional setting. A new engineer joins the company and begins the process of becoming productive. They start by checking out the codebase and understanding its structure and conventions. The engineer then reviews and interprets feature requests or bug reports, creates a high-level design for the new feature or bug fix, and moves on to implementation. This involves creating new files and functions as needed, writing code to implement the design, optimizing the code for performance and readability, and often using tools like ChatGPT to assist with code generation. Finally, the engineer tests their changes to ensure functionality and catch any bugs, then commits their changes and pushes them to the shared repository.
Over the last year or two, we've seen a significant acceleration in software development productivity due to AI tools like ChatGPT. Many engineers now leverage these AI assistants to generate code snippets, debug issues, and even write entire functions given clear specifications. This has dramatically reduced the time spent on routine coding tasks. However, these tools still require human engineers to orchestrate the entire development process, from understanding requirements to integrating the AI-generated code into the larger system.
At NonBioS, we're taking AI assistance in software development to the next level by automating the entire engineering process. Users interact with a NonBioS AI engineer through a chat interface, much like they would communicate with a human colleague. When given a task, the NonBioS AI engineer can perform all the steps traditionally done by a human engineer. It accesses and analyzes the existing codebase directly from the source control system, interprets feature requests and bug reports, designs solutions at a high level, implements the solutions by writing code, optimizes the code, and conducts testing to ensure the changes work as intended.
The NonBioS AI engineer will be able to check out code, makes changes, and checks them back in directly, without the need for a traditional IDE. Our AI system will continuously learn from the codebases it interacts with, improving its understanding of different programming paradigms, design patterns, and company-specific conventions. It can collaborate with human engineers, taking direction, answering questions, and providing explanations for its decisions.
Related Posts
