Skip to content

BYOK — Bring Your Own Key

Pro and Enterprise customers supply their own LLM API key. Helm routes AI requests through your key — we never see your traffic, never mark up your compute costs, and you choose the model.

Privacy — Your diagnostic commands, ticket contents, and M365 data never pass through Helm’s servers. They go directly from your Helm instance to your chosen LLM provider.

Cost control — You pay your LLM provider directly at their posted rates. Helm doesn’t add a markup. For most MSPs, the AI cost per technician per month is $5–15.

Model choice — You’re not locked into one model. Choose based on cost, capability, or data residency requirements.

  • OpenRouter — Recommended. Single key, access to Claude, GPT-4o, Gemini, Mistral, and others. Pay per token at market rates.
  • Anthropic — Direct Claude API access.
  • OpenAI — Direct GPT-4 API access.
  • Azure OpenAI — For organizations with Microsoft data residency requirements.
  • Any OpenAI-compatible endpoint — Ollama, LM Studio, or other self-hosted models.

OpenRouter is the simplest setup:

  1. Create an OpenRouter account
  2. Add a credit card and set a spending limit
  3. Generate an API key
  4. Paste it into Helm’s settings

You’ll have access to every major model. Default recommendation: Claude Sonnet for the best balance of capability and cost.

Free tier users receive 200,000 tokens per month included — no API key required. This covers light use (a few hours of troubleshooting per month). Pro and Enterprise users bring their own key for unlimited use.