Anthropic says it made Claude.ai and the Claude desktop app's core UX ~3x faster in August using an internal model to ship improvements "in a two-week sprint"

Users of Claude.ai and the desktop app will experience significantly faster response times and a more fluid interaction. Developers can now leverage AI-assisted performance tuning to achieve rapid improvements in their own applications. The success of this sprint signals a new capability for AI to autonomously identify and resolve software performance issues at scale.

Anthropic announced that it significantly improved the performance of its Claude.ai website and desktop application during August. The company executed a two-week sprint focused on core user experiences, achieving an average speed increase of approximately 3x across key journeys. Specific improvements included reducing the time to a typeable page on a fresh load of claude.ai from 3.1 seconds to 0.55 seconds, and decreasing the time to start a new Claude Code session from 0.8 seconds to 0.3 seconds. These enhancements were driven by an internal research model, Claude Tag, which identified bottlenecks, built benchmarks, and proposed solutions. The process involved extensive collaboration within a Slack channel, with Claude actively participating in monitoring deploys, assessing telemetry, and suggesting optimizations, leading to the implementation of thousands of changes without negative customer impact.

The rapid, AI-driven performance optimization at Anthropic demonstrates a tangible shift towards agentic development workflows, where AI models actively participate in the software development lifecycle beyond mere code generation. This approach allows for hyper-iterative improvements, compressing what would typically be months of engineering work into weeks or even days. The success here suggests that companies can achieve substantial user experience gains by integrating AI tools not just for creative tasks but for systematic, data-driven problem-solving in performance engineering.

This advancement has implications for the broader AI market by showcasing a practical application of AI in improving core product performance, a crucial differentiator for user retention and satisfaction. It also sets a new benchmark for AI-assisted development, potentially influencing how other AI companies and software developers approach optimization and feature delivery. The ability to rapidly tune user-facing applications using AI could accelerate the pace of innovation across the industry, making complex performance gains more accessible.