Aithy
A personal AI agent that can work safely on your machine, remember useful context, and keep its data under your control.
Last verified:
What is Aithy?
Aithy is a private local AI runtime designed for state-of-the-art agent research, local inference, sandboxed tools, durable memory, and complete data control. It is a personal AI agent that runs safely on your machine, built on Bun, DSPy, and RLM (Recursive Language Model) to make small models perform exceptionally well.
Key features include search and scraping capabilities, solid memory systems for remembering useful context over time, reusable skills that agents can call upon, sandboxed execution for safe code running, clean worker architecture, and base model support. The agent can help with projects, maintain persistent memory, and execute code in a secure sandboxed environment while keeping all data under your control.
Aithy is ideal for researchers conducting agent research, developers building local AI applications, privacy-conscious users who want AI without cloud dependency, and anyone who needs an AI agent that works offline with complete data sovereignty. It runs locally on your hardware without requiring internet connectivity after setup.
Aithy pricing
Pricing model: Freemium
Aithy is open source under Apache License 2.0 and free to use. There are no monthly fees or paid plans. The only costs are one-time hardware costs for running local AI (computer with adequate RAM/CPU/GPU) and optionally purchasing API credits from model providers if using BYOK (Bring Your Own Keys) for certain models. All features including agents, memory, skills, and sandboxed execution are completely free.
Aithy pros
- Runs completely locally on your machine
- All data stays under your control
- Works offline without internet connection
- Sandboxed execution for safe code running
- Durable memory system for context retention
- Reusable skills architecture for extensibility
- Built on Bun for fast performance
- Uses DSPy + RLM for small model optimization
- Search and scraping capabilities built-in
- Clean worker architecture for maintainability
- No ongoing monthly fees after setup
- Complete privacy - no data leaves your machine
- Supports state-of-the-art agent research
- Permission-based sandboxing for JS and VM runtimes
- Open source under Apache License 2.0
Aithy cons
- Requires local hardware with sufficient resources
- Setup complexity for local AI infrastructure
- May need GPU for optimal performance
- Smaller models may have limitations vs cloud APIs
- Self-hosted maintenance responsibility
- Limited to models you can run locally
- No cloud backup for memories
- Initial hardware investment required
Frequently asked questions about Aithy
What is Aithy?
Aithy is a private local AI runtime and personal AI agent that runs safely on your machine. It can work with state-of-the-art agent research, perform local inference, use sandboxed tools, maintain durable memory, and keep all your data under your complete control. It is built on Bun, DSPy, and RLM to make small models perform exceptionally well.
Is Aithy free to use?
Yes, Aithy is completely free and open source under Apache License 2.0. There are no monthly fees, subscription costs, or paid plans. The only costs are one-time hardware costs for running local AI on your machine and optionally API credits from model providers if you choose to use BYOK for certain models.
How does Aithy's memory system work?
Aithy has a durable memory system that remembers useful context over time. The memory persists locally on your machine, allowing the agent to maintain context across sessions and projects. All memories stay under your control and never leave your device.
What is sandboxed execution in Aithy?
Sandboxed execution allows Aithy to run code safely in isolated environments. It sandboxes both JS and VM runtimes with permission-based access, ensuring that any code the agent executes cannot harm your system or access data it shouldn't. This provides security while still allowing the agent to perform useful tasks.
What are reusable skills in Aithy?
Reusable skills are pre-built capabilities that Aithy agents can call upon when needed. Skills declare what capabilities they need, and the runtime enforces these requirements. This modular architecture allows agents to extend their functionality by loading skills as needed, similar to how coding agents read skills from the filesystem.
Does Aithy work offline?
Yes, Aithy works completely offline after initial setup. Since it runs locally on your machine with local inference, it does not require an internet connection to function. This provides privacy, cost savings, and the ability to use AI even without internet access.
What models does Aithy support?
Aithy supports various local models that can run on your hardware. It is optimized to make small models perform well using DSPy and RLM. You can use different model families depending on your hardware capabilities, and optionally use BYOK (Bring Your Own Keys) to connect to external model providers when needed.
What hardware do I need to run Aithy?
Aithy runs on consumer-grade hardware, though having a GPU speeds up performance significantly. You need a computer with adequate RAM and storage for running local LLMs. A GPU is not strictly required but recommended for better performance, especially with larger models.
How does Aithy handle search and scraping?
Aithy has built-in search and scraping capabilities that allow the agent to gather information from the web when needed. These features work within the sandboxed environment, ensuring safe execution while providing the agent with access to current information.
Who is Aithy designed for?
Aithy is designed for researchers conducting state-of-the-art agent research, developers building local AI applications, privacy-conscious users who want AI without cloud dependency, and anyone needing an AI agent that works offline with complete data sovereignty. It is particularly suitable for those who want to run AI on their own hardware without ongoing costs.