Uber Cuts AI Costs by 34% While Scaling Agent Usage 9.4x
dosco · x · 2026-08-28
Uber shares how it scaled AI agent usage across its software factory while stabilizing costs. With over 70% of PRs attributed to agents and 3,600+ skills built, Uber optimized its infrastructure to drive down the cost per 1,000 requests by 34% and cost per session by 52%, despite a 9.4x growth in weekly agentic requests. The post details their approach to vendor-neutral managed agents, SWE benchmarking, and prompt optimization.
More from coding & agent
- Sentry engineer: Clients like Codex are unreliable, so servers must self-heal — zeeg · 2026-08-28
- Trying free Claude Code: proxy routing burns quota instantly, Kaggle self-hosting too slow — phantom_root · 2026-08-28
- Agents Feared Failing METR's Auto-Scorer for 'Cheating' to Grab the Flag — BLUECOW009 · 2026-08-28
- Detecting Unicode stealth attacks: A security scanner for MCP Servers — Ecstatic_Muffin8929 · 2026-08-28
- Linear migrates to StyleX: 1,000 PRs to prepare for AI agents — talkaboutdesign · 2026-08-28
- Foremerge: Open-source protocol to coordinate parallel coding agents above Git — ShiftTechnical · 2026-08-28