AI for IT Operations, Without Giving Up Control

AI can transform IT work, but many organisations cannot send system data to any cloud. Cyana lets you choose where AI processing happens, including models running on your own hardware.

Start Free Trial View Pricing

Why private AI matters for IT

IT operations data includes machine names, users, software, logs and configuration. For regulated, isolated or privacy-conscious organisations, sending that to a public AI service may not be acceptable. Private AI keeps processing inside infrastructure you trust.

Your AI, your infrastructure

Cyana is designed around choice: of model, of provider and of where processing happens.

Choose your provider

Use Anthropic Claude, OpenAI, Google Gemini, Azure OpenAI, Mistral or Cohere, depending on your requirements.

Run models on-premise

Point Cyana at an OpenAI-compatible endpoint serving a tool-capable model on your own hardware.

Self-hosted end to end

Combine a self-hosted SimpleHelp server with a local model so neither sessions nor prompts leave your network.

Same controls as SimpleHelp

Cyana follows your technician permissions, machine scopes and review settings.

AI assistance in an isolated network

A research organisation runs SimpleHelp inside an isolated network. It hosts an open-weight language model on a local GPU server and connects Cyana to it. Technicians use natural-language fleet queries and in-session diagnostics without any data leaving the site.

Cyana Command in the SimpleHelp Technician Console generating a machine fleet overview report from a plain-language request. — Ask Cyana a question about your machines and get a structured answer.

Make the trial match your real work

Start with a few test devices and a workflow your team knows well.

Choose a provider or local model that meets your policies.
Connect Cyana to it and run a read-only query.
Review which features use external services in your deployment.

Learn more about Cyana →

Your questions, answered

Which local models work with Cyana?

Models served through an OpenAI-compatible API that support tool use. Capability varies by model, so test with your typical workloads.

Is a local model as capable as a cloud model?

It depends on the model and hardware. Many routine fleet queries work well with smaller models; complex tasks may benefit from larger ones.

Do I pay SimpleHelp for AI usage?

Cyana is licensed per machine. Model usage is paid to your provider, or runs on your own hardware.

See all solutions →

Use AI on your terms

Put SimpleHelp to work with your own machines and technicians. Every feature, free for 30 days.

Start Free Trial