Breaking

Anthropic Says Claude.ai Became About 3x Faster After Two-Week AI-Assisted Sprint

September 23, 2026 at 15:18 EDT

AI Agents
Software Dev & Coding

Anthropic says it made the core Claude.ai and Claude desktop experience about 3x faster during a two-week August sprint, using Claude itself to measure bottlenecks, build benchmarks, propose fixes and monitor deployments. The company published the engineering account on September 23, saying the work covered journeys that represent 95% of user activity and landed more than 3,000 changes without a customer-facing incident or rollback.

Sep 23, 2026 · Anthropic

Claude.ai got faster by measuring every wait

A two-week sprint used Claude to find bottlenecks, build benchmarks, ship fixes and watch deploys across the core Claude web and desktop experience.

Core experience speedup

95%

User activity covered

3,000+

Changes merged

Customer-facing incidents

Fresh load wait collapsed

At the 75th percentile, time to a typeable Claude.ai page fell from 3.1 seconds to 0.55 seconds.

3.1s

Before

0.55s

After

The optimization loop

Measure a journey

→

Build or find a benchmark

→

Ship behind guardrails

→

Watch field data

Why teams will study it

The method turns vague slowness into measurable targets, PRs, deployment checks and ratcheting tests.

What still needs work

The process was human-steered, and Anthropic says tail latency, other journeys and very long conversations remain targets.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.