Loading…
Loading…

NALA — Network Agent Logic Architecture
For thirty years the editor was the center of software: you and a file, one line at a time. When agents arrived, the first instinct was to bolt them onto the editor we already had. Cursor forked VS Code; the agent became a guest in a house built for a single typist.
We think the house itself has to change. When you are conducting a fleet of agents rather than editing one file, the workspace should be built for that from the ground up: durable terminal panes, Tasks, messages, and approvals as first-class citizens. That is NALA. Not a fork. An agentic IDE, or ADE (the term people are starting to use for this new category), built from the ground up: a base layer for agentic engineering, available free, and it stays free. It extends the tools you already have, and runs entirely on your own machine.
The name says the thesis: naturally autonomous LLM agents. Agents should behave more like people working together than software following scripts: each with its own role, skills, memory, and tools, dividing the work, asking for help, reviewing each other. Cooperative, transparent, provider-agnostic. You describe the goal; the system figures out how.
The team photo: one founder, a fleet of agents, all working inside nala itself.
nala is a Windows-first base layer built on open foundations and shaped by a single founder conducting a fleet of AI coding agents inside the product itself. Thousands of tests keep the fleet honest. No investors to thank yet, no advisory board to parade: just the software, shipped free for the people who use it.
The longer road is an operating layer for your whole machine: open nala, speak or type, and conduct agents to do the work. We are building the backend infrastructure to keep it updated and dependable for the long haul while the product remains local-first, using the models and agents you choose.
“After a few days of dogfooding I stopped toggling between terminals and virtual desktops entirely. I command whole teams of agents and put agents in charge of other agents. It genuinely feels like superpowers, and that feeling is what we're shipping.”
The product is free: download it, run it locally, bring the agents and provider tools you already use. No Nala account or checkout is required today.
We are moving toward open-sourcing the codebase so people who care about the craft can read it, improve it, and ship with us. Access to the repo will go through an approval process, not a public free-for-all.
That is intentional. The last few years of public open source have been flooded with drive-by pull requests and machine-generated noise that burn maintainers and bury real work. We would rather know who is at the door, keep the signal high, and treat contribution like membership in a small workshop, not a spam folder.
If you want source access or a path to contribute, write us. Tell us who you are and what you want to build. Serious people get a serious answer.
We're rebuilding the IDE for the agentic era. Follow our progress, get product updates, and be the first to know what ships next.

Brand new, local-first, and free. Be among the first to conduct a fleet of agents from a workspace built for it.
The Nerve Center for Agentic Engineering