Plane AI scores 165/165 on real workspace eval: 18s median latency, $0.10 per answer
JosephJacks_ · x · 2026-11-04
Plane's team significantly upgraded the harness powering its AI, cutting median response time by over 20%, and shared a full real-world eval:
- Setup: a workspace for a fictional company with 64 people, 2,560 work items, and 10 projects
- Prompts: from simple ("Which urgent Payments items are overdue, and who owns each?") to complex root-cause analysis ("What caused the ledger double-posting incident, and which post-mortem actions remain open?")
- Results: 55 questions × 3 runs, Claude Sonnet 5.5 at high effort answered 165/165 correctly (100%)
- Cost & speed: 18s median time to answer, $0.10 median cost per answer (the scatter plot shows most answers under 20s/$0.10, with a few complex ones taking 80+s and $0.70+)
The team believes Plane AI will eventually replace search and much of the UI in the product: users ask, and the AI finds or does the work.
More from coding & agent
- Zero-shot mobile manipulation demo stacks Dream-1 policy, MCP planner and ROS nav — scott_e_reed · 2026-11-04
- Anthropic launches Claude auto-connect beta, letting admins onboard whole orgs to MCP plugins — TansuYegen · 2026-11-04
- Coding agents build baselines for 36 top-journal AI methods, matching ~95% of scores — owl_posting · 2026-11-04
- Postman desktop auto-installs Codex/Claude plugins, tries to trust its own startup hook — jedisct1 · 2026-11-04
- Oracle's agentic AI product lead: enterprise AI roadmaps run on outdated use-case lists — akumar · 2026-11-04
- Sentdex: vibe coding is real, OpenHands shows dev may not involve reading code — Sentdex · 2026-11-04