The complete answer — what we're building, why, and where it's going.
Aevnis manages and runs your life. Four AI modules cover every domain that matters: what you build, what you know and track, who you communicate with, and what you remember. Each module works by natural language or conventional controls — same data, both ways. Your machine. Your data.
The premise is simple: AI has made it genuinely possible for one person to do what previously required a team of five or ten. A solo developer can now build, ship, and maintain products at a pace that used to require an engineering department. A freelancer can take on client engagements that once needed an agency. The ceiling for what one person can do has moved dramatically upward.
But the tooling hasn't kept up. The AI tools that exist are fragmented, cloud-dependent, and designed for enterprise compliance budgets. They're great at individual tasks — autocomplete here, a chatbot there — but they don't add up to a coherent working environment. The operator is still left holding everything together manually.
Aevnis is the desktop environment built to change that. It brings together an AI development agent (Corvus), an AI operations assistant (Wren), persistent cross-project memory, configurable autonomy, and direct access to the model APIs you choose — all in a single desktop application that runs on your machine, with your data, under your control.
Independent operators using AI today face three structural problems that no existing tool has solved.
You have ChatGPT open in one tab, GitHub Copilot in your editor, Notion AI in another. None of them know what the others are doing. You spend more time coordinating your AI tools than doing the actual work.
Your prompts, your projects, your IP — all flowing through someone else's servers. Enterprise AI tools are designed for enterprise compliance budgets. Independent operators get none of the privacy guarantees.
Every AI conversation starts from zero. No memory of your projects, your codebase, your preferences. You're the only one holding it all together — which defeats the purpose of having AI in the first place.
Four principles shape how Aevnis is designed. Every decision traces back to at least one of them.
Aevnis Shell runs on your machine. Your projects, conversation history, and settings live on your computer. The AI models — Claude, Gemini, Ollama — are called directly using your own API keys. There is no Aevnis server in the middle storing your prompts or reading your files. What you work on stays with you.
Corvus knows your project structure, your git history, your running processes, your previous conversations. Memory persists across sessions and across projects. You don't repeat yourself. You don't re-explain your codebase every morning. The AI carries the context you've already established.
From "ask before every action" (Level 1) to "run until done" (Level 5), you decide how much agency Corvus gets per conversation. High-stakes refactors get Level 2. Routine scaffolding gets Level 4. You set the dial. Corvus respects it. Nothing irreversible happens without your say.
Aevnis is not designed for a 200-person engineering team. It's not designed for a CTO with a six-figure software budget. It's designed for one person who needs to operate at scale — to ship code, produce documents, manage projects, and run operations without hiring anyone to help.
Aevnis is built on four modules that together form a complete operating layer. Corvus handles technical work. Wren manages your financial life, records, and everyday operations — money, documents, anything that needs tracking. Kestrel handles communications. Owl holds the memory that makes everything coherent over time. They share context, run in the same desktop shell, and are built around a single principle: the operator should be deciding what matters — Aevnis handles the execution.
Corvus is an AI development agent that lives in your desktop. You describe what you need — a new feature, a bug fix, a refactor, a shell script — and Corvus plans, executes, and reports back. It reads and writes files on your system, runs terminal commands, manages git, searches the web, calls external APIs, and remembers everything across projects and sessions.
Corvus supports Claude (Anthropic), Gemini (Google), and local Ollama models. Autonomy is configurable per task: set it low for risky operations, high for routine work. Available now at $14.99/mo.
Wren is your complete life administration layer. It tracks what you owe and what you're owed, reads your payslips, manages invoices and subscriptions, monitors insurance and investments, and maintains your financial history — all in one place. Beyond money: shopping lists, poker logs, receipts, documents, and any records you need organised. Talk to it naturally or use structured panels — same data, both ways.
$14.99/mo
Kestrel is your AI communications layer. It triages your inbox, drafts contextual replies, schedules meetings, and manages follow-ups. Never miss an important message or let a thread go cold. Kestrel understands who matters and what needs a response right now.
$14.99/mo
Owl is your AI second brain. It captures notes, articles, ideas, and research — then surfaces exactly the right knowledge exactly when you need it. Over time it builds a personal knowledge graph that grows smarter with every thing you save, connecting ideas you’d never have linked yourself.
$14.99/mo
The long-term vision is a unified Dashboard where all four modules report to a Master AI with full context across your entire operation — code, finances, communications, and knowledge. Think of it as an AI chief of staff: it keeps everything coordinated, surfaces what needs attention, delegates to the right module, and maintains a coherent picture of where your work and life stand at any moment.
Ultimately, Aevnis is building toward an AI-native operating system — a Linux-based OS purpose-built for operators who want AI deeply integrated at every layer: the shell, the file system, the window manager, the networking stack. Not a chatbot bolted on top of Windows. An operating system where AI is the operating layer — where every file operation, every terminal command, every workflow is AI-aware by default.
AevnisOS is the five-year vision. Aevnis Shell is Phase 1 — it ships now, works today, and proves the model that the full OS will be built on.
Aevnis is not for everyone. It's for a specific type of person: someone working independently who needs to operate at scale without the infrastructure of a company behind them.
You're building a product alone or with a tiny team. Corvus is your developer, your devops engineer, your code reviewer. You stay in the product and strategy layer while Corvus handles execution. You ship faster than teams three times your size because you're not managing humans — you're managing context.
You take on client projects across multiple stacks and languages. Corvus helps you move fast on unfamiliar codebases, drafts technical documentation via Wren, writes client proposals, and keeps project context separate and private per engagement. You deliver more, bill more, and spend less time on the work that doesn't require you specifically.
You run a small service business, agency, or consultancy. You need to write code, manage projects, produce documents, and run operations — all without hiring a team. Aevnis is your force multiplier: the combination of Corvus and Wren means you can take on work that previously required at least two or three people, and deliver it at a standard you're proud of.
For those who want to know how it works — here is the honest technical picture.
Tauri v2 (Rust)
The desktop wrapper. Tauri gives Aevnis Shell a native window on Windows, macOS, and Linux with a minimal Rust runtime. It's lighter than Electron, faster to start, and has a much smaller attack surface. The Rust layer handles sidecar process management, IPC, and system-level calls.
Python sidecar
The AI engine runs as a Python subprocess communicating via JSON-RPC 2.0 over stdin/stdout. This keeps the AI logic separate from the Rust layer, easy to update without rebuilding the full Tauri app, and auditable — the Python code is readable by any developer who wants to inspect it.
Vanilla HTML/CSS/JS
The frontend is plain HTML, CSS, and JavaScript — no React, no build step, no node_modules. This was a deliberate choice: the interface should be fully readable, fast to modify, and not dependent on a framework ecosystem that changes every two years.
FastAPI + SQLite
Auth, subscriptions, and licence verification run on a FastAPI server with a SQLite database, hosted on a VPS in Sydney, Australia. The backend stores only what it needs: email, hashed password, subscription status. It never sees your prompts, your files, or your API keys.
Direct API calls
Your Claude, Gemini, or Ollama API keys go directly from your machine to the respective AI provider. Anthropic's servers see your prompt. Google's servers see your prompt. Aevnis servers never do. There is no man-in-the-middle on the AI calls.
GitHub Releases (CI/CD)
Every release is built automatically by GitHub Actions across all four target platforms (Windows x64, macOS Intel, macOS Apple Silicon, Linux x64). The same commit produces all binaries. Releases are signed and distributed via GitHub Releases with auto-update support built into the app.
Aevnis Co. is an independent software company founded in Burradoo, NSW, Australia. It is a one-person operation — which means the product is built by its own target user. Every feature that ships is a feature the founder needed. There is no product team deciding what users want. There is one person who is the user, building the tool he wants to exist.
The company is bootstrapped and early-stage. The alpha version of Aevnis Shell is live, working, and already being used for real work. The roadmap is not a marketing document — it's a sequence of things that need to exist for the product to be complete, in the order they make sense to build.
Not a professional programmer by background. Uses AI as his primary development tool — which is, perhaps, the best possible qualification for building a product designed to help others do the same. Based in the Southern Highlands of New South Wales, Australia.
Aevnis exists because the tools the founder wanted didn't exist. The goal was a desktop AI agent that ran locally, remembered context across sessions, could operate at configurable levels of autonomy, and didn't route everything through a cloud platform. Nothing on the market was that. So it got built.
The company name was coined for this project. It doesn't mean anything in any existing language — it's a name that belongs entirely to Aevnis.
This is an honest picture of where the product is and where it's going. No hype, no "coming Q3" vagueness — just phases, in order, with what each one actually delivers.
The alpha is live. Corvus, Wren, and Owl are working. Corvus handles code and automation. Wren manages your financial life and records. Owl captures notes, memories, bookmarks, and reminders. 7-day free trial, no credit card required. Available on Windows, macOS, and Linux.
Wren and Owl are live. Wren handles financial and life records. Owl is the persistent memory layer across all modules. Kestrel — email, calendar, and communications — is next.
The product reaches its designed state: Corvus stable, Wren live, Dashboard working, all three talking to each other with shared context. This is the version that no longer needs the "alpha" qualifier. Polished, fast, reliable enough to be someone's primary working environment.
Development begins on the operating system proper: a Linux-based OS where AI is integrated at every layer. The shell knows about your projects. The file manager knows about your agents. The window manager understands your workflow. This is not a distribution with a chatbot installed — it's an OS designed from the ground up for AI-native operation.
The complete vision: a Master AI that coordinates Corvus, Wren, and any future modules with full operational context — code, documents, finances, communications, calendar. The operator sets direction. The AI stack executes. The gap between "what I want to happen" and "what is happening" approaches zero.
7-day free trial. No credit card required. All platforms.