GitHub CEO Thomas Dohmke on Building Copilot, and the the Future of Software Development
Training Data
Agents Transform Tasks into Solutions
Agents and copilots operate on a similar principle, utilizing models to accomplish tasks efficiently. An agent executes a specific task by interacting with the model, whereas a copilot functions as a multifaceted agent, integrating various features to assist users. For example, an auto-completion feature acts as an agent by analyzing user input and context to generate intelligent predictions. The evolution of agents will lead to capabilities that autonomously handle more complex tasks, such as the anticipated auto-fix feature. This feature aims to streamline the process of addressing security vulnerabilities in code by automatically detecting and suggesting fixes for issues like SQL injection or cross-site scripting upon a pull request submission.
