AWS recently introduced Lambda Durable Functions, a new approach for serverless workflows that directly challenges the established Step Functions service. The core problem is cost: Step Functions charges per state transition, meaning you pay even when workflows are simply waiting for human input or external events. Durable Functions solve this by suspending execution during wait times, resulting in zero charges for idle periods.

Developers write workflows in standard TypeScript or Python using familiar `await` syntax. AWS handles state persistence and resumption automatically. This code-based approach contrasts sharply with Step Functions' JSON state machines.

The cost savings are dramatic for long-running processes; a document approval workflow processing 50,000 documents monthly might cost $9.90 with Step Functions but only $0.03 with Durable Functions—a 99.7% reduction. While Durable Functions excel at human-in-the-loop patterns and scheduled batch jobs, Step Functions retain advantages in visual editing, massive parallel processing, and direct AWS service integrations. For developers building new workflows with significant wait times, Durable Functions offer a compelling, cost-effective alternative.