Throughput Scaling for High-Volume AI Agent Deployments
Most AI agent pilots fail at scale due to infrastructure, not model capability.
Mirela Ostrowski
Senior Editor, Infrastructure & Search
Mirela Ostrowski
Mirela Ostrowski spent a decade as a distributed systems engineer at mid-size cloud infrastructure firms before pivoting to technical journalism in 2014. She focuses on search architecture, indexing pipelines, and the infrastructure stacks that underpin production AI deployments.
4 stories
Retrieval Latency Budgets in Multi-Step Agentic Workflows
Break retrieval into stages to find where multi-hop agents actually lose time.
Mirela Ostrowski
Web Retrieval API Design Principles for LLMs
Retrieval, not the model, is where most AI agents fail in production.
Mirela Ostrowski
Factuality Decay in LLMs Without Live Web Grounding
Models trained on static data drift away from current facts without live data connections.
Mirela Ostrowski