I've watched enough coding agents work to understand why software velocity is exploding: code work is made of small, reversible, compounding steps. Data work has none of this. Too much of it happens in large, throw-away iterations: copies, scratch buckets, one-off datasets that never add up to anything. The part where agents come in, of courseWith coding agents, generating experiments is no longer the bottleneck — risk is. What makes a fleet productive is that each agent's work lands as a transaction the others can see, build on, and walk back.