We have added an incredibly powerful tool to iii: observers and they solve an all-too-common problem in software: Want a ping whenever an order over $500 comes in? Usually you'd edit the order provider, rebuild, test, and redeploy. What about when a request errors? Same flow, and same maintenance burden. Eventually you'd integrate other system(s) to handle events, error reporting, and plenty more. This is one of many growing pains every project goes through. We wanted to avoid it from the start. So now in iii 0.25 you can observe any function inputs, outputs, and errors. In any language. From any service. Want to know when anything happens? Define an observer in your business logic, and leave your other services alone. And we improved our execution overhead (1.9 μs -> 1.3 μs).

We were getting tired of massive inference bills. Most of it was agents rereading context, calling --help again and retrying payloads they guessed wrong, so we’ve been squeezing that out of iii. A feature on DeepSeek Flash now costs about ten cents, and on Claude is 24% cheaper. iii.dev/blog/coding-agent-co…

Linear, Jira, Trello, GitHub Projects, Multica, every coding agent. Same kanban board, rebuilt every single time. We recorded one of our team members building a PR review app in the iii ADE, start to finish, in about an hour. The first prompt asked for org and repo selection, a PR list with filters, per-file review marks in SQLite, agent questions on selected lines, and local worktrees. 5 minutes later it was running, using the database and github services that were already there. Twenty minutes in it needed a kanban board. It found one, added it, and it was working immediately. No integration, no different languages, no figuring out HOW to integrate it. Over that hour the agent wrote 19 files, +3,044 lines, every single one focused on the task at hand: the PR review app. Nothing else needed to change. The kanban service was still pinned at 0.1.20. Sergio built the board, the whole team extended it, and every app that composes it gets ongoing improvements automatically. Need a database? iii trigger compose::add worker=database. A browser, a queue, a cron, an HTTP endpoint? Add them. 101 workers on our registry.

Shipped today: tmux in iii. It even tells you if an agent is waiting on your input. Claude Code, Codex, OpenCode or Gemini CLI, doesn't matter, they all work.

Introducing Kits, self-contained collections of services, integrations, agents, and skills that install and run with a single command. No more needing to assemble your tool, agent, and skill stack yourself. No more stacks that sort of work together. No more out of date and mismatched tools. Almost ready. Comment "kits" down below to get notified when we release it.

There are hundreds of companies building cloud agents, plus the big labs. We're working one layer down, so your whole backend, regular services and agents, runs as one system on iii. We'll offer cloud agents too, we're just not going to build our own.

Simulated comparisons of models and harnesses are useful, but in AI’s stochastic world, real-world tests always tell the true story.

We made updating your entire service architecture as easy as runing npm update. Complete with versioning, failure recovery, and dependency chains.

Keeping up with your community is a full-time job. There's Github Issues & PRs, Discord discussions, X and LinkedIn posts, and a whole lot more going on. So we built a single unified dashboard that replaces and ranks many disparate sources of data. With iii. In a matter of hours. But what's even better than this tool is that it is also a production-ready system. Shipped without encountering a single bug. Thanks to iii 100% of our effort was spent on features.

We just shipped support for TypeSafe, Laya, SemIf, and Decider. Thanks to iii's unique architecture each is a drop-in replacement for the other so try them out in the backend, frontend, or our harness and let us know what you think.

We've built an agent-assisted onboarding that takes you from zero to a working production-ready project while teaching you the basics of iii. Get iii from our website and try it out yourself with `iii project init --learn-iii`.

Same model. Two harnesses. Wildly different builds.

Luca is doing things in our harness that we've never seen done before. Read his thread to learn why 👇

Adding Jev to our tool decision making shaved an entire minute off of our agent guided onboarding. We're at 7 minutes and targeting 5 minutes to onboard. Where do you think the biggest Jev gains would be next?

How can two agents with the same model, same prompt, and same budget produce drastically different results? Simple, build a system that frees up the agent to solve the problem rather than build the infrastructure to solve the problem. Jump to 0:22 to see the difference. This small demo represents how future systems will be built: realtime, and language-agnostic. With composability, observability, discoverability, extensibility, and reactivity as basic properties. Each of these properties contributed to the iii agent's performance.

Recovery is often overlooked in harnesses. If a process fails it just waits for the agent to realize and to try to restart it. There's no restart policy other than what the agent decided in the moment. It's often difficult to tell what is running, what has been orphaned. A productive agent session quickly devolves into staring at a process monitor. We flipped that on its head by making the agent part of the architecture.