Follow the latest coverage, related explainers and connected technology stories.
GitHub explains how to create interactive work interfaces inside the GitHub Copilot app by describing a workflow in natural language, then modifying the interface and using it alongside the agent. These interfaces can be used for Kanban boards, issue triage, checklists, and dashboards.
Microsoft has unveiled a redesign of Copilot around three core capabilities: Home, which unifies chat and delegated work; Code, which creates applications and automations using natural language; and Autopilot, which proactively performs ongoing tasks. This is accompanied by running Office inside Copilot, a Managed Runtime architecture, and new tools for managing enterprises’ AI spending.
OpenAI launched the GPT-6 Sol and GPT-6 Luna models for use in ChatGPT, Codex, and GitHub Copilot, cutting API prices in half compared with the prices of comparable GPT-5.6 models. The models deliver improved performance in some programming and alignment tests, but safety evaluations reveal continuing challenges in monitoring and responding to malicious instructions.
GitHub explains how it rebuilt the pull request interface in the GitHub Copilot application to handle diffs exceeding one million lines and more than 400 inline comments. The solution separated the layout of predetermined code lines from dynamic comment blocks, using incremental measurement and correction that preserves the user’s position.
GitHub explains the migration of the shared runtime for Copilot products from TypeScript and Node.js to more than 800,000 lines of Rust, through 128 pull requests and gradual replacements within the main branch. The experience reveals how AI agents were used to accelerate a project that would previously have required an entire team for one or two years.