Frequently asked questions
What is Helix?
Helix is an open-source AI platform that lets you run powerful language models on your own infrastructure with complete control over your data and models.
What makes Helix different?
We allow you to safely bring LLMs into your business by running models in your environment. This means you have zero data leakage risk and complete control over your AI infrastructure.
What is Helix Code?
Helix Code lets you run fleets of AI coding agents that work autonomously on software development tasks. Key features:
- GPU-accelerated desktops - Each agent runs in a real Linux desktop environment with full IDE access
- Live streaming - Watch agents work in real-time through your browser
- Spec-first development - Agents write design documents before code, ensuring thoughtful implementation
- Kanban orchestration - Manage dozens of agents from a single dashboard
- Private and secure - All code stays on your infrastructure, never sent to external services
Helix Code is ideal for teams that want to accelerate development while keeping proprietary code secure.
Do I need to use the Helix UI to interact with my agents?
No. You can interact with Helix agents through any integration connected to our Streaming API. This includes Teams, Slack, Azure DevOps, and many other integrations.
How hard is it to get started?
See getting started! Installation typically takes 15-30 minutes with our automated installer.
Do I need a GPU?
No, a GPU is optional. You can use Helix in two ways:
Without a GPU: Connect to external AI providers (OpenAI, Anthropic, Together) and route your queries through Helix for logging, access control, and app building.
With a GPU: Run open-source models locally for complete data privacy. NVIDIA GPUs with 24GB+ VRAM are recommended.
What open source models can I use with Helix?
Almost any popular model will work with Helix: versions of Llama, Deepseek, Phi, Qwen, Aya, and more. For image generation, Flux is supported.
Can I integrate my models with my own apps or business apps?
Yes, we expose industry-standard OpenAI compatible APIs. Please see the API documentation.
We also provide “App” primitives to make it even easier for you to create AI-powered applications.
What is RAG (Retrieval Augmented Generation)?
RAG lets you “chat with your documents.” Upload PDFs, Word docs, or other files, and the AI will answer questions based on their content. This is useful for creating chatbots that know about your company’s specific information.
How much does Helix cost?
Helix is open source and free to self-host. Your main costs will be:
- Compute: Server or cloud instance costs
- GPU (optional): If running local models, GPU rental or hardware
- AI API costs (optional): If using external providers like OpenAI
Is my data safe with Helix?
When self-hosted, your data never leaves your infrastructure. Conversations, uploaded documents, and fine-tuned models all stay on your servers. This is ideal for sensitive business data or compliance requirements.
What are Helix Apps?
Helix Apps let you create custom AI assistants. You can:
- Define a system prompt (personality/instructions)
- Connect knowledge bases (documents the AI can reference)
- Add tools (let the AI take actions like searching the web or calling APIs)
No coding required for basic apps!
Can I run Helix on Kubernetes?
Yes! We provide Helm charts for Kubernetes deployments. See the private deployment guide for details.
How do I get help?
- Check this FAQ and our documentation
- Visit our GitHub Issues
- Join the Discord community
- Email us at [email protected]