Iii

If you’re a backend engineer tired of juggling queues, crons, and streams, iii’s three-primitive model might be the relief you need. The promise of adding a new capability with `iii worker add` is compelling—it treats services like npm packages, which fits how many of us already think. We’d reach for this when we want to prototype a polyglot event-driven system quickly, especially one that mixes Node.js, Rust, Python, and Go. But it’s not for everyone. iii is self-hosted only; there’s no managed cloud yet, so you’re responsible for deployment, scaling, and uptime. If your team prefers graphical configuration over code-defined primitives, or you need no-code integration flows, this isn’t the tool. Enterprises requiring SOC2 or HIPAA out of the box will need to look elsewhere, as iii is early-stage and doesn’t offer those compliance certifications. Compared to alternatives like Temporal or Dapr, iii’s single-protocol approach over WebSocket is its key differentiator. Temporal relies on a dedicated workflow engine, and Dapr uses sidecars for cross-language communication. iii avoids both by making cross-language calls native, which reduces operational complexity. But Temporal is more battle-tested in production, so if you need mature workflow guarantees, weigh that trade-off. In practice, we’ve seen iii shine for agent builders. Because agents are just workers, you don’t need a separate harness—your agentic system is part of the runtime, which can save tokens and reduce moving parts. The recent recoverable triggers and per-subscriber retry tuning (v0.22.0/0.22.1) make it more resilient for real-world deployments. Watch out for the learning curve. While the primitives are simple, reasoning about distributed systems still requires care, especially with durability and