Yana AI – AI Ecosystem & Development Environment

Yana is an AI ecosystem and software development tool for macOS, focusing on three core areas: connecting multiple AI providers, controlling the execution process, and maintaining user decision rights.

Instead of relying on a single AI model, Yana allows connection to multiple providers through APIs, selecting the right model for each task and integrating tools within the same ecosystem.

1. Yana App – AI Application

A native macOS application built with SwiftUI, providing an interactive environment for working with AI.

Multi-provider AI: Connect multiple AI models through the APIs of different providers.

AI Chat: Manage conversations, save histories, and quote responses for ongoing exchanges or rebuttals.

Long-term Memory: Manage long-term memory, supporting data export and backup.

Permission Management: Manage permissions through Auto Mode with temporary rights and Custom Mode for customization.

Plugin & MCP Integration: Access tool ecosystems like Claude Plugins, MCP Registry, and ChatGPT Apps List.

Built-in Terminal: Integrate a command line environment.

Update Management: Check for updates to Yana App, Yana Runtime, and Yana AI.

2. Yana Studio – Development Environment

Yana Studio is a software development environment customized from Code-OSS, serving programming and source code management tasks.

The Studio operates as a standalone application, with its own window and environment, not embedded in the Yana App. The two applications are distributed together in a single installation package.

3. Yana Runtime – Execution Control Layer

Developed in Rust, Yana Runtime is an independent runtime layer responsible for managing policies and controlling the execution of AI actions.

The Runtime uses three control outcomes:

Allow: Permits actions according to policy.

HumanApprovalRequired: Requires user approval.

Deny: Denies disallowed actions.

Instead of merely processing requests and generating responses, the Runtime evaluates actions according to policies, determining the range of actions that AI is permitted to perform and when human approval is needed.

Key Differences

Tools like Cursor, Windsurf, and integrated AI IDEs focus on integrating AI into the software development process. Yana approaches from a broader system perspective: combining AI applications, a development environment, and a separate Runtime layer to manage execution permissions.

Three design principles:

Model-neutral: Does not lock users into a single AI provider. Multiple models can be selected according to needs.

Runtime-level Execution Control: Separates the AI interaction layer from the execution control layer. The Runtime evaluates actions according to policy.

Human-governed AI: AI can assist and perform tasks, but users determine the scope of authority and approve necessary actions.

For example, in a software development workflow, a user might select one model for code analysis, another model for writing documentation, and another model for supporting test case creation. The models are used within the same ecosystem, while the Runtime maintains a layer of action control.

Yana is not merely aimed at becoming another AI IDE. The project focuses on the ability to connect multiple AIs, tools, and workflows within a controlled architecture.

Target Audience

Yana is aimed at programmers, independent developers, software engineers, and those who frequently use multiple AI services, especially those who want to integrate AI into workflows while still needing control over system access and actions.

Platform and Getting Started

Operating System: macOS 14 and above.

Hardware: Apple Silicon from M1.

Installation: Download the application from GitHub Releases.

AI Configuration: Users connect the desired provider's API key themselves.

Store API Key: macOS Keychain.

Yana is an open-source project, developed with the goal of building a flexible, scalable AI ecosystem that places human control at its core.

AI can act. You stay in control.