The credit-pool split Anthropic announced June 1 went live on schedule. Programmatic usage — anything hitting the API or SDK directly — no longer competes with Claude.ai seat consumption for the same token budget. For teams that mix human-facing and automated workloads under one account, this changes how billing is tracked and potentially how cost-control policies need to be written. It's a sensible separation that's been overdue, but it does mean existing billing dashboards and alerting thresholds may need recalibration. Check your usage breakdown in the Anthropic console to confirm which pool your workloads are landing in.