VIOLET AI PLATFORM
Secure by default. Flexible by design.
A controlled AI platform developed in Sweden and operated on Swedish infrastructure.
THE CORE
Build AI on your terms. Not the vendor's.
Control over operations, data, and models.
Choose how and where your AI environment is run. Full control over data storage, access, model selection, and logging with support for your security and regulatory compliance requirements.
Control over where your data is located
Access on your terms
You choose the AI models
Storage and retention according to your requirements
Encrypted all the way
Full traceability with an audit log
AI connected to your business
Connect Violet to your existing documents, data sources, and business systems. Permissions and integrations ensure that the right users get access to the right information.
Connect your existing sources
Keep your permissions
Integrate via API and MCP
AI tailored to role and process
Answer from your own knowledge
Traceable to the source
INTEGRATIONS
Connecting to your world.
Violet can be connected to internal systems, external data sources, and existing workflows via API, MCP, and webhooks.
Entra ID
SSO · Roles · Users
MCP
Model Context Protocol
REST API
Full API access
SharePoint
Document sync
Teams
Notifications · Bots
ABOUT THE PLATFORM
Your questions.
Which AI models can we use, and can we switch?
The platform is not locked to a single vendor. You choose the model per use case and can switch as needs change, without rebuilding assistants, integrations, or permissions. For sensitive information, models can be run within your own infrastructure.
Which systems can Violet be connected to?
Violet connects to documents, databases, and business systems via API, MCP, and webhooks. Common integrations include Entra ID for identity and permissions, SharePoint for documents, and Teams for use in daily work. Deeper integrations with, for example, ERP and CRM are built based on your specific conditions.
Does the AI retain our existing permissions?
Yes. Access is controlled by your existing authorization structure, which means a user only gets answers based on information that their role already grants access to. You also decide which data sources each assistant is allowed to use.
How do we know that the answer is correct?
The answers are retrieved from your own sources and show where the information comes from, down to the file, page, and paragraph. The user can therefore verify the basis directly. This does not eliminate the need to review results in decisions that have consequences, but it makes verification possible.
What is the difference between an assistant and an agent?
An assistant answers questions and helps in a dialogue. An agent performs a multi-step task, scheduled or event-driven, and reports the result. Both work within the permissions and rules you have set up.





