ServiceNow CoreAI builds AutoSynthData to turn agent failures into training tasks

ServiceNow CoreAI published AutoSynthData, a method that converts a target agent's failures plus a stronger teacher model's successful trajectories into verified synthetic training tasks for the same enterprise environment. The approach lets production agent teams continuously generate environment-valid tasks with verifiers and train on the gaps that still break their agent, replacing hand-curated eval failures with reusable training curricula.

Hugging Face https://huggingface.co/blog/ServiceNow-AI/autosynthdata ServiceNow CoreAI builds AutoSynthData to turn agent failures into training tasks Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated. AutoSynthData turns a target agent’s failures plus a stronger teacher model’s successful trajectories into verified synthetic training tasks for the same enterprise environment. For production agent teams, the practical shift is from hand-curating eval failures into reusable training curricula: you can continuously generate environment-valid tasks with verifiers, train on the gaps that still break your agent, and move the frontier as capabilities improve. Mistral Large quota or rate limit — check usage and plan. Original headline: AutoSynthData: Generating Training Data for Enterprise Agents