Updated October 25, 2026
· ~2 min read
Every system eventually needs a workflow that survives failures: an order that waits days for payment, a data pipeline with retries, a long-running AI agent that must resume after a crash. Three approaches dominate on AWS. Step Functions defines workflows as state machines and runs them as a service. Temporal lets you write them as ordinary code (durable execution, where the workflow's state survives process crashes) and runs either on your own cluster or as Temporal Cloud. Argo Workflows runs each step as a container on Kubernetes and is popular for data and ML pipelines.
Step Functions Standard costs $0.025 per 1,000 state transitions after a free 4,000 a month; Express costs $1.00 per million requests plus duration.
Step Functions Standard is cheap until a workflow does many small steps at high volume: a ten-step workflow run a million times a month is ten million transitions, about $250. Express Workflows are far cheaper for high-volume, short workflows. Temporal's costs are your cluster and database or Temporal Cloud's actions-based pricing; Argo's are the containers each step runs. The usual mistake is a chatty Standard workflow — loops of tiny steps — that would cost a tenth as much as Express or as one Lambda doing the loop.
Question 1 of 2
Is the workflow mostly calls to AWS services?
Use Step Functions for AWS-centric orchestration — calling Lambda, ECS tasks, Bedrock and dozens of AWS APIs directly with retries and error handling, and nothing to run. Use Temporal when workflows are complex enough that you want them as code with tests, versioning and local debugging — long-running business processes and AI agent loops fit it especially well — choosing Temporal Cloud unless you have a reason to run the cluster. Use Argo Workflows for container-per-step batch, data and ML pipelines on a Kubernetes cluster you already run.
Serverless Migration
Move from expensive always-on compute to serverless architecture: reduce costs, eliminate patching, and scale automatically.
AWS Architecture Review
A focused, senior-level review of your current AWS architecture, with a prioritized list of what to fix and why.
Production GenAI Platform on AWS
Deploy secure, scalable AI systems using AWS-native GenAI services. RAG, agent orchestration, and guardrails: production-ready.
We work with startups and growing companies on AWS architecture decisions. Tell us about your situation and we will recommend the right approach.
As of October 25, 2026. Provided as is, for general information, and accurate to the best of our knowledge on the date shown. Not legal, compliance, tax, financial or other professional advice. Prices, limits and vendor services change; check current vendor documentation before relying on anything here. Terms of use.