A general coding agent can write a training script that looks like it works, until it blows up a large model's training budget on GPUs and pathological runs. The solution to this is a specialized harness with a focus on training AI models, with a repository of individually tested and vetted training recipes for every model and use case, developed by engineers who lived the problem in post-training.
We gave an AI agent a raw base model, one H100 and 10 hours. It took Qwen3-1.7B-Base from about 10% to 75.9% on GSM8K, 1,001 of 1,319 test problems, scored by the benchmark's own unmodified evaluator in a fresh container. No human touched the run.
The agent was GPT-5.6 Sol at max reasoning, driven by our harness. It chose the data, wrote the training code, ran supervised fine-tuning and then RL, checked its own work for contamination, and shipped a merged model. It finished in 7h59m, with 241 tool calls and about 18.5M tokens.
We're building ashr, a platform that helps teams build, evaluate, and monitor AI agents, catching errors before production. The same failure patterns kept showing up across our users' repos: broken tool calls, mismanaged context, and agents begging for tools that they didn't have access to.