Full control over your AI applications
FinOps, Governance, and Compliance for your AI agents and applications. Keep costs in view at all times with Token Control and manage access to your AI models efficiently and securely. In real time, with data residency in the EU.
Seamless integration with leading AI providers
FinOps – Keeping track of costs
Requests that would exceed a limit are stopped before costs are incurred.
Managing your AI budgets & limits
Manage the AI budgets of your user groups and users centrally. Define budget limits for departments, applications, projects, or individual users. Distribute your AI budgets according to the needs of your company or organization and avoid unplanned costs.
Monitor & analyze your AI costs
Monitor your AI costs and the budget status of your users clearly and in real time. Get monthly cost reports for your internal billing processes. Analyze your AI consumption based on your user groups, users, and budgets.
Plug-and-play integration with existing AI solutions
Integrate Token Control into your AI solution through our transparent REST API-based approach. Simple, with no additional development effort. Also compatible with your existing AI platforms and chats.
Suad Wolgram, Product Manager, white duck GmbH
Governance – Access under control
Directly controllable and documented in an auditable way.
Organization of your business entities
Manage your business entities centrally by clearly structuring user groups, departments, cost centers, or solutions such as AI agents, chats, and applications. Manage access rights and budgets in a targeted manner to ensure transparency, efficiency, and control over your AI models.
API key management for your AI solutions
Create and manage dedicated API keys for your AI solutions to control access individually and ensure security. Enable targeted use by specific applications, projects, or user groups and maintain an overview of resource control for your AI models.
Control access to your LLM models
Monitor and manage access to your LLM models with clearly defined TPM (tokens per minute) limits per application. Set individual usage limits for applications, teams, or projects to ensure regulated resource usage and guarantee optimal performance of your AI models.
Nico Meisenzahl, COO, white duck GmbH
Compliance – Meet regulatory requirements
Token Control provides the operational foundation for implementing these requirements.
Compliance Dashboard
The standalone Compliance Dashboard displays the regional model distribution, an overview of content filters (showing blocked, filtered, and allowed requests per level and category), as well as recent activities.
Full Traceability
Every request is traceably documented with a Correlation ID, including token usage, model deployment, timestamps, and status information. Your content remains private, prompts and responses are not stored.
Data storage and processing in the EU
All data is processed and stored within the EU, fully within the EU Data Boundary. In the optional private model, Token Control runs directly in your Azure subscription, aligned with your existing security and governance policies.
Markus Sümmchen, CEO, white duck GmbH
Start now
SaaS
For a quick and easy start to your AI management.
Fully managed solution
Integrate with your existing AI solution
Pay-as-you-Go
Data residency in the EU
Privat
For your AI solution with complete data sovereignty.
Deployed in your own Azure environment
Isolated data storage
Tailored to your individual needs
Operated by us
Partners
For your market as our partner.
Dedicated multi-tenant environment
Your individual branding
Integrated with your solutions
Tailored to your business model
With NOVA—our customized AI assistant—we can use AI securely and in compliance with data protection regulations within our company. Thanks to its seamless integration into our corporate environment, intuitive operation, and tailored responses. Working with white duck is always professional and straightforward, meaning short communication channels, quick feedback, and a strong team with diverse AI and cloud native expertise.
white duck GmbH as our innovation partner!
They advise us, provide us with some concepts and a turnkey solution with full cost control for our computer science, business informatics and artificial intelligence students. This enables us to adapt teaching concepts and also provide students with lightweight programming interfaces. The first projects for industrial partners are already underway.
FAQ
How does Token Control support EU AI Act requirements?
Token Control provides the technical foundations relevant for implementing EU AI Act requirements: traceable usage via Correlation ID, regional model distribution in the Compliance Dashboard, a content filter overview, dedicated API keys per use case, and auditable metadata. The legal assessment of your AI application remains your responsibility — Token Control provides the data foundation.
How long does implementation take?
With the SaaS model, you are typically up and running within a few days. A technical integration via the compatible interfaces requires no code changes to your existing AI applications. For more complex setups, we recommend a scoping workshop.
Which AI models are supported?
Token Control supports all common LLM models, including:
Azure AI Foundry:
All common OpenAI models, Microsoft models (Phi, model-router), Mistral AI, Meta, DeepSeek, xAI, Nvidia, as well as many Hugging Face and open-source models.
(Azure) OpenAI:
GPT (Chat, Mini, Nano), GPT-OSS, GPT-Image, Text-embedding
Google Gemini:
Gemini (Pro, Flash, Flash-lite)
Mistral AI:
Large, Medium, Small and Document AI
Meta Llama:
Llama 4 (Maverick, Scout, Behemoth)
In addition, Token Control supports all open-source LLM models with an OpenAI-compatible API.
How does the integration work?
Token Control can be easily integrated into your existing AI solutions and enables simple, efficient implementation. Without any additional development effort.
Seamless integration:
All requests sent to your AI models are processed by Token Control and forwarded unchanged. This preserves the full functionality of your existing systems.
No code changes required:
AI solutions that already use Azure OpenAI / Azure AI Foundry or other supported models can be operated directly with Token Control without any changes to the code. This saves valuable time and resources and enables quick and easy implementation.
Flexible model management and scaling:
Token Control supports the scaling of AI models and the management of multiple model deployments. This allows you to run different models in parallel and respond flexibly to the requirements of your applications, projects, or teams.
Are third-party integrations supported?
Token Control can be easily integrated into third-party solutions. Thanks to its plug-and-play architecture, it can be quickly and easily integrated into existing AI chats, chatbots, and tools such as GitHub Copilot Chat. You can seamlessly integrate Token Control into your existing systems without any complex adjustments or additional development work.
Integration is achieved through the use of dedicated API keys, which ensure a secure and controlled connection between Token Control and your third-party solutions. These API keys allow you to precisely control access to your AI models and efficiently monitor usage. This gives you full control over your AI resources at all times and enables you to implement governance and cost policies in a targeted manner.
Are web search and deep search functions supported?
Token Control offers comprehensive support for grounding and deep search through seamless integration with leading AI agent frameworks such as FLOCK (our open-source AI agent framework), Langgraph, LangChain, Semantic Kernel, and Autogen. These frameworks enable you to connect your AI models to external data sources, delivering more accurate and contextually relevant results.
Support for web search is a key part of our roadmap and has the highest priority. If you have specific requirements or use cases, please contact us to discuss your needs and work together to develop the optimal solution.
Is RAG (Retrieval-Augmented Generation) and Grounding supported?
Token Control fully and directly supports RAG (Retrieval-Augmented Generation) and Grounding “out-of-the-box.” This allows you to seamlessly connect your AI models to external knowledge sources or databases to generate more accurate and context-aware responses. Thanks to its simple and efficient implementation, you can use these technologies without additional development effort, significantly increasing the performance of your AI solutions.
How does the Microsoft Entra ID integration work?
Token Control supports management via Microsoft Entra ID groups (formerly Azure AD). You can use M365 and security groups to map API keys and business identities such as user groups, departments, or cost centers. This enables centralized and structured control of access to your AI resources based on existing organizational structures.
How does Token Control relate to API gateways?
Token Control can be easily integrated with an API gateway (e.g., Azure API Management). While Token Control solves organizational challenges such as the management and reporting of API keys, user groups, departments, or cost centers, the API gateway handles technical aspects such as routing and load balancing. Together, they offer a comprehensive solution that addresses both organizational and technical hurdles.
Is Token Control GDPR compliant?
Token Control is fully GDPR compliant and was developed with a clear focus on data protection and data security. All data is processed and hosted exclusively within the EU, using the proven infrastructure of Microsoft Azure. By using Azure's EU Data Boundary, we ensure that all data flows and storage comply with the strict requirements of the European General Data Protection Regulation (GDPR).
Who is behind Token Control?
Token Control is developed and operated by white duck GmbH. white duck has been operating as a Microsoft Solution Partner based in Rosenheim since 2012.
