Designing latency budgets for real-time AI
A practical way to break an AI response into network, queue, preprocessing, inference, and application budgets before choosing where it should run.
Read articleNew: Designing latency budgets for real-time AIRead the article
NYR blog
Local AI
engineering.
Architecture, infrastructure, and hard-won lessons for teams building AI that has to perform beyond a distant cloud.
FeaturedFrom the NYR team
Latest articles
5 articles
01Edge AI fundamentalsWhat is edge AI? A practical guide to local inference
Learn how edge AI runs inference closer to users, devices, and data, and when local deployment is the right architecture for an AI product.
02ReliabilityDesigning AI that keeps working when the cloud does not
How to define local autonomy, degraded modes, recovery, and data synchronization for AI products deployed at customer sites.
03ArchitectureEdge AI vs. cloud AI: how to choose the right architecture
Compare edge AI and cloud AI across latency, privacy, availability, cost, operations, and personalization to choose the right deployment model.
04InfrastructureA control plane for heterogeneous edge compute
The abstractions an AI platform needs to place, update, observe, and recover workloads across unlike customer-site infrastructure.
05Deployment guideHow to plan a multi-site AI deployment
A practical framework for taking an AI workload from one successful pilot to a reliable, private, locally tuned rollout across customer sites.
New to local AI?
Start with the fundamentals.
Field Notes organizes our durable explainers and deployment frameworks into one practical starting point.
Browse field notesBuild closer