Today for AI
HOT RADAR
agentHEAT 14.0°

AI CURATED HOT STORY · 11/15/2026

Northwestern and Google Unveil Self-Improving Agent Harness Method

1 reports archived1 independent sourcesupdated 11/15/2026, 08:08:32
Synthesis & Latest Updates
1 Sources Cross-Validated

A Northwestern and Google paper introduces Harness2, which lets agents self-improve their harness (prompts, tools, skills, memory) on new tasks without answer keys by test-running proposed edits and keeping only those that help. A single improvement step lifted GLM-5 by 11.5 points on WorkBuddy-Bench, surpassing Gemini 3.5 Flash on its stock harness, while unvalidated harness edits dropped coding scores from 73.4 to 63.2.

LATEST/A Northwestern and Google paper introduces Harness2, which lets agents self-improve their harness (prompts, tools, skills, memory) on new tasks without answer keys by test-running proposed edits and keeping only those that help. A single improvement step lifted GLM-5 by 11.5 points on WorkBuddy-Bench, surpassing Gemini 3.5 Flash on its stock harness, while unvalidated harness edits dropped coding scores from 73.4 to 63.2.

Hourly heat curve

4 fully observed hours
Now
14.0
Peak
14.909:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

Coverage timeline

Total 1 reports · Latest first
  1. Rohan Paul (@rohanpaul_ai)T2·78 pts

    Northwestern and Google Unveil Self-Improving Agent Harness Method

    • Harness2 self-improves per task without answer keys by test-running proposed harness edits and keeping effective ones
    • One improvement step raised GLM-5 by 11.5 points on WorkBuddy-Bench, beating Gemini 3.5 Flash on its stock harness
    • Unvalidated harness edits cut coding scores from 73.4 to 63.2, showing real-run verification is essential