Task model recipes

One recipe = one scoped job your agent already runs. We train a small open-weight model on your labels; your harness keeps calling a model — just not the expensive flaky one.

• Decision / risk classifier task modelTrain a scoped decision/risk task model on your outcomes and put it under your agent — beat frontier cost and flakiness on one labeled job.min 500 rows · lora_sft
• Fraud / payments risk task modelReplace frontier LLM fraud calls with a small open-weight task model trained on your payment outcomes — cheaper, measurable, callable by your agent.min 1,000 rows · grpo_rlvr