Turning fuzzy judgments into classifiers with Opus 5.5 agents — no manual labeling needed
ivan_bezdomny · x · 2026-10-04
The author shares a hands-on experiment using AI agents (Astra and Opus 5.5) to convert fuzzy judgments into classification problems, using name merge/conflict resolution as an example. Key points:
- No manual example tagging required — just guidance and a natural dataset
- Opus automatically mines hard examples and compares AUC, effectively doing the ML workflow itself
- Results reportedly beat both rule-based systems and direct LLM-as-judge approaches
He suggests inspecting failures or having Opus surface the highest-entropy examples for review, and notes it would make a great student project — though Opus does all the work.
More from coding & agent
- Leave comments on Paper, agents turn them into motion code: 'design to get a better prompt' — davidhoang · 2026-10-04
- A few nights with Claude and three.js: a water sim in the browser that still wants a duck — minchoi · 2026-10-04
- "It Sends a System Message to My Agent Fleet": Reboots Now Brief AI Agents Like Engineers — kevinnbass · 2026-10-04
- Copilot fixes Surface Book 3 camera driver on Linux in two hours — marian_nmt · 2026-10-04
- New Opus went from flawless to average, says dev, but still beats Opus 5 — haider1 · 2026-10-04
- Redditor burns through entire ChatGPT Go plan in one day building a Minecraft mod — SuperJogosDBZ · 2026-10-04