Boreon MCP Layer · Tableau Server & Cloud
Bring your own LLM. Keep your own keys.
The Boreon MCP Layer lets you choose the model for every task over Tableau Server or Tableau Cloud. Self-hosted installs keep the keys entirely yours.

Three ways to plug in a model.
Use one model for everything, or a different one for each kind of work.
-
Anthropic
Claude
Claude Sonnet, Opus or Haiku, chosen per task.
-
OpenAI
GPT
GPT models through the OpenAI-compatible slot.
-
Your endpoint
Any
Any OpenAI-compatible https endpoint, from open models such as Qwen and Gemma hosted by Infomaniak in Geneva to NVIDIA Nemotron.
Each task on its own model.
The platform routes four kinds of work separately. Keys are encrypted at rest, and every routing change is audited.
| ClaudeAnthropic | GPTOpenAI | AnyYour endpoint | |
|---|---|---|---|
| Chat | |||
| Drafts | |||
| Escalation | |||
| Classify |
Your Tableau. Your rules.
It reads Tableau Server or Tableau Cloud read-only, with SSO, MFA, a full audit trail and a web application firewall in front.

-
Self-hosted or Swiss-hosted
A signed offline bundle for your own cloud, or hosted by us in Switzerland.
-
Read-only by default
The one write is a workbook you ask to publish.
-
SSO, MFA and audit
Entra ID or any OIDC provider, with every action on the record.
-
A WAF at the front door
OWASP Coraza with the Core Rule Set.
Built by a Salesforce Partner
Ask your own Tableau, live.
Bring one question from your Tableau Server or Tableau Cloud site. The Boreon MCP Layer will answer it in line, on the model you choose, and that question becomes the first thing we set up.