pneuma
software that materializes from intent
see it in action
how it works
you describe what you want. it appears.
every program is generated, compiled, and executed at runtime.
what is pneuma
pneuma is an AI-native computing environment. Instead of installing pre-built applications, you describe what you want in plain language — a calculator, a dashboard, a game — and a working program is generated, compiled to WebAssembly, and executed on your GPU in seconds. Every agent runs in a sandboxed thread with capability-based permissions. There are no dependencies, no supply chains, and no install steps. The software exists only for as long as you need it.
run anything
why pneuma
instant software
Describe what you want. A working program appears in seconds — generated, compiled, and running natively on your machine.
sandboxed by default
Every agent runs in a WebAssembly sandbox with a 64 MB memory limit. No filesystem access, no network access — unless you approve it.
no dependencies
Each agent is a single self-contained Rust program. No npm, no build tools, no supply chain. The source code is inspectable and readable.
native performance
Compiled to WebAssembly and rendered with wgpu (Vulkan/Metal/DX12). Pure Rust desktop app — no Electron, no browser engine.
frequently asked questions
What is pneuma?
pneuma is an AI-native operating system where software is generated from natural language prompts. You describe what you want, and a working program is written in Rust, compiled to WebAssembly, and executed in a sandboxed thread with GPU rendering — all in a few seconds. There are no pre-installed applications.
How does pneuma work?
You type a plain language prompt describing what you want (e.g. "make a calculator"). An LLM writes a Rust program, the code is compiled to WebAssembly, and it runs in a sandboxed thread with GPU rendering via wgpu. If compilation fails, the system automatically retries with error context. The entire pipeline takes a few seconds.
Is pneuma free?
Yes. The free tier gives you 10 generations per day and 30 per month. The Pro tier costs $5/month and includes 100 daily generations, agent editing and refinement, the ability to publish to the agent store, and priority generation queue.
What platforms does pneuma support?
pneuma is available for Windows, macOS (Intel and Apple Silicon), and Linux. It is a native desktop application written entirely in Rust — no Electron or browser engine.
Is pneuma safe to use?
Every agent runs inside a WebAssembly sandbox with a 64MB memory limit and capability-based permissions. Agents cannot access your filesystem, network, or hardware directly — all I/O goes through capability-checked system agents. The source code of every agent is inspectable.
How is pneuma different from ChatGPT or AI coding assistants?
ChatGPT generates text. Coding assistants like Copilot help developers write code faster. pneuma generates running software — when you ask for a calculator, you get a real compiled program rendering on your GPU, not a code snippet. The target user is anyone who can describe what they want, not just programmers.
What can I build with pneuma?
Anything that fits on a screen: calculators, dashboards, games, clocks, note-taking apps, data visualizers, API monitors, chat interfaces, crypto trackers, todo lists, and more. Agents can make HTTP requests, store data persistently, play audio, and communicate with each other via IPC.