#ai
3 posts tagged ai.
- Rate limits are product decisions, and AI agents just became your biggest customersAPI gateways have two very different kinds of limits: ones that protect your systems and ones that define what a customer bought. In 2026, agents that never sleep are testing both.
- The GPU line item is the new EC2 line itemAI workloads moved the biggest number on the bill. The playbook that tamed EC2 still applies, with two new metrics: GPU utilization and cost per request.
- SLOs for features that answer differently every timeLLM-backed features break the usual assumptions: slow by design, a dependency you do not run, and "success" that a status code cannot see. What to measure instead.