Key insight: tool discipline > scale. Instead of complex multi-table training, we reinforced reliable tool use on simple queries — and saw transfer to 12-step reasoning tasks (59.7% Pass@1).

→ View original post on X — @snorkelai