Extella

Extella is a desktop AI execution platform built by Chariot Technologies Lab., Inc. that turns plain-language descriptions into working automations running locally on your device. Unlike chat-based AI tools that reset after each session, Extella accumulates knowledge, preferences, and reusable pipelines — making the system progressively more capable over time. It is available for macOS, Windows, and Linux (desktop and CLI worker variants).

What It Is

Extella positions itself as a "self-evolving agentic platform" built around three core primitives: Experts, Concepts, and Rules. Experts are isolated execution containers — reusable scripts or multi-step pipelines that run locally on the user's machine in any language or runtime (Python, Node.js, Bash, R, Perl). Concepts are persistent semantic memories retrieved by vector similarity, so the right technical context fires automatically on relevant queries. Rules are a dynamic behavioral layer that rewrites how the underlying model reasons, adapting from user corrections and persisting across all future sessions. Together, these three primitives let users describe a task once and re-run it indefinitely without re-explaining context.

How the Expert System Works

Experts come in four levels of complexity:

• Simple Experts — single-purpose tools (e.g., send_telegram_message, convert_pdf_to_text)
• Complex Experts — multi-step pipelines wrapped into one callable unit (e.g., decompile a binary to pseudocode, generate a 3D model from a photo)
• Nested Experts — orchestrators that chain multiple sub-Experts into end-to-end business workflows
• Integration Experts — wrappers around specific technologies such as external APIs, ML frameworks, CLI tools, databases, and Python libraries

No coding is required to create Experts; users describe the desired outcome in plain language and Extella generates and saves the Expert to a personal library.

Architecture and Local Execution

Extella runs Expert code directly on the user's device rather than in a cloud sandbox, giving it access to local files, applications, and hardware. API keys and secrets are stored in encrypted, isolated storage and are never hardcoded into automations. The platform sits above the user's existing stack and orchestrates connections to services including Claude, ChatGPT, Gemini, local LLMs, GitHub, Telegram, WhatsApp, Slack, Gmail, Notion, Google Sheets, CRMs, and databases. A CLI Worker variant for Linux (x86_64 and aarch64) enables headless or server-side execution.

Use Cases and Target Audience

The platform targets knowledge workers, operations and RevOps teams, marketers, developers, and individuals who want a personal automation OS. Documented use cases include:

• Research that compiles briefs, sources, and structured reports while learning house style
• Operations pipelines connecting CRMs, spreadsheets, drives, and APIs without glue code
• Content production from outline to publish-ready asset with persistent tone and brand rules
• Engineering scaffolding — spinning up internal tools, scripts, and integrations from a single instruction
• Personal task management where recurring routines and decisions are remembered and acted upon
• Data analysis where users describe a question and receive both the answer and the artifact

Update: Version 1.0.71

The homepage and download links reference version 1.0.71 as the current release for macOS (Apple Silicon), Windows 10/11, and Linux desktop builds. A separate CLI Worker at version 1.0.0 is available for Linux x86_64 and aarch64. The product is actively distributed and described as free to install, with a token-based model for ongoing usage. A macOS Toolbar extension built by the Extella team is also available via a shell installer script, demonstrating active first-party tooling development on top of the base platform.