Inngest

Inngest is a platform designed for building and orchestrating durable workflows, background jobs, AI agents, and serverless endpoints without requiring developers to manage infrastructure. At its core, Inngest provides "durable functions" that replace traditional queues, state management systems, and scheduling services. Developers write code as regular functions but wrap them in Steps — a primitive that automatically handles retries, recovery, and execution flow. This means that if a function fails mid‑execution due to a network error, a service timeout, or any transient issue, Inngest will automatically retry the failed step from its last checkpoint, preserving all prior state and ensuring eventual completion. No manual error handling, no custom retry logic, no external state stores. The system is event‑driven: developers can trigger functions via HTTP requests, scheduled cron expressions, or other events within the system. Inngest manages the entire lifecycle — from event ingestion to execution, throttling, concurrency control, and observability. It runs on a distributed, horizontally scalable backend that handles millions of operations per day, yet developers never see a queue, a worker pool, or a database. The platform is "infraless" — meaning the orchestration engine does not dictate how code is written. Functions can be deployed on any runtime (Node.js, Python, Go, etc.), and can be integrated with existing frameworks like Next.js, Express, FastAPI, or plain serverless functions. A key use case is AI workflows. Teams building LLM‑powered applications often need to chain multiple API calls (e.g., prompt engineering, model inference, vector database lookups, post‑processing). Inngest allows each step to be defined as a separate Step with built‑in retries and timeou