# Temporal and AWS: A partnership built for production-grade AI

DevFeed: [Temporal and AWS: A partnership built for production-grade AI](<https://devfeed.tech/articles/temporal-and-aws-a-partnership-built-for-production-grade-ai-86243.md>)

Original publisher: [Read original article](<https://temporal.io/blog/temporal-and-aws-a-partnership-built-for-production-grade-ai>)

Author: Jay Sivachelvan

Published: 2026-10-13T00:00:00Z

Content type: article

Language: en

Sources: [Temporal](<https://devfeed.tech/sources/temporal-blog.md>)

Topics: [Amazon Bedrock](<https://devfeed.tech/topics/amazon-bedrock.md>), [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [Amazon Elastic Kubernetes Service](<https://devfeed.tech/topics/amazon-elastic-kubernetes-service.md>)

Tags: [agentcore](<https://devfeed.tech/tags/agentcore.md>), [ai](<https://devfeed.tech/tags/ai.md>), [aws](<https://devfeed.tech/tags/aws.md>), [partnership](<https://devfeed.tech/tags/partnership.md>), [product-news](<https://devfeed.tech/tags/product-news.md>), [production](<https://devfeed.tech/tags/production.md>), [strands-agents](<https://devfeed.tech/tags/strands-agents.md>), [workflow](<https://devfeed.tech/tags/workflow.md>)

## AI overview

Temporal and AWS are expanding their partnership around production AI systems. The article describes Temporal integrations with the Strands Agents SDK and Amazon Bedrock AgentCore, plus preview compute options that let teams run durable workflows without keeping worker fleets active when idle. It also reports a joint customer pattern that reduced backend infrastructure costs by 66% while processing millions of events per month.

## Source excerpt

The AI stack on AWS has never been stronger. Amazon Bedrock gives teams access to leading models and the Strands Agents SDK makes agents easier to build. Amazon Bedrock AgentCore provides a managed runtime for operating them. Each layer does its job, and one production question still spans all of them: will the application actually finish what it started when the model calls time out, tools fail, infrastructure restarts, or a workflow waits for hours or days for a person to respond?