At Aistatine, we build on-device AI models so your code, prompts, and development workflow can stay on your device.
We started Aistatine because we were frustrated. The best AI coding tools all had the same caveat buried in the terms of service — your prompts and code are sent to servers you don't control, used to train models you didn't consent to, and stored for longer than you'd like.
Aistatine’s public product is @code: a terminal-first tool that installs a local model and runs through a local inference server. The documented workflow supports macOS, Windows, and Linux, and can work offline after its initial model download.
We build for solo developers, startup teams, and organizations assessing private AI coding workflows. We publish product details in the How It Works guide and keep the installer public so users can inspect the setup process.
Aistatine was founded in 2026 around a practical question: can an AI coding workflow be useful without making a hosted model the default destination for a developer’s repository? Our answer is to make local operation the starting point. The public installer downloads the binary and model to the developer’s machine, and the local setup can continue to work without a network connection once it is installed.
That approach comes with responsibilities. Developers still need to review diffs, run tests, protect devices, and follow their organization’s source-control and deployment processes. We do not present local AI as a substitute for engineering judgment. We build it as an option for teams that need to assess privacy, cost control, and operational boundaries alongside coding speed.
We will add company registration details, formal licensing information, and fuller founder biographies when they are available for publication. Until then, this page links directly to the founders’ public profiles and the @code installation repository so visitors can inspect the product information that is public today.
For product requirements, release questions, or a conversation about local AI deployment, contact Aistatine. We would rather answer a specific question with current evidence than make broad claims that a team cannot validate.
We don't collect telemetry, training data, or usage statistics. Privacy isn't a setting you turn on — it's the baseline.
We build tools that augment your judgment, not replace it. Aistatine shows its work and waits for your approval.
Every action the agent takes is logged and visible. You always know what Aistatine did and why.
Internet access is optional. Aistatine runs fully offline so you can build on a plane, in a SCIF, or anywhere in between.
Built by founders who believe privacy-first AI is the only AI worth shipping.
We're happy to walk you through Aistatine and answer any questions you have.
