your agent's traces only cover what already broke. synthetic data should not imitate them it should extrapolate and generate the failures that haven't happened yet. early results on real production failure diagnosis: - trained on ZERO real traces, it matches a model trained on real expert traces (82% vs 82%) - few-shot post-training stops being a lottery (every run 80–86%) - the model sees failure types no expert ever labelled introducing edgecase🐵a platform to extrapolate your agent's traces into training data. recipe and code are fully open source.

check out: app.bandr.ai

had so much traffic, it went down now back again :D

@CompleteSkeptic @madiator @arimorcos Would love your feedback guys!

and it's live on the platform: bring your traces, get training data back. app.bandr.ai recipe and code are open source in bandits: github.com/bandr-ai/bandits

the expensive part of post-training was never compute. it was expert time. a few real traces + edgecase data: 85% ~3× more real expert traces: 87% nearly the same result for a fraction of the labelling, and ~$5 to generate the data. ~$14 in total compute