Which LLMs are supported?
Currently, you can use the Co-Analyst feature in ChatGPT, Claude, and Microsoft Copilot. Or via direct access in the minubo chatbot to our optimally selected LLM providers with EU hosting and zero data retention.
Is my data secure? Will it be used to train public AI models? What about GDPR compliance?
To ensure your data is secure, it's crucial that the LLM you connect to the minubo MCP infrastructure is a Business or Enterprise account. This is the only way to prevent your data from being used for training purposes. We also recommend carefully reviewing your LLM's configuration and, if in doubt, consulting your legal advisor.
When using the minubo chat, you're fully protected: We only use LLM providers with EU hosting and zero-data retention. This means your data is neither used for training purposes nor stored by the LLM provider for any other purpose.
Furthermore, our co-analyst is configured in both the chat and the MCP setup to prevent any personal data from being shared with the LLM providers. Therefore, you can use AI analytics in full compliance with the EU GDPR.
Do I need programming skills?
No. You ask your questions in normal, everyday language – just as you would speak to a colleague. The co-analyst understands natural language and translates it into the necessary analyses. No SQL, no Python, no technical setup required.
What are minubo Playbooks?
Playbooks are structured analysis frameworks for complex use cases such as campaign evaluation or forecasting. They incorporate decades of e-commerce expertise: which questions to ask, which comparisons are important, and which outliers are typical. You don't need to know what the perfect specification for a complex analysis looks like; with the help of the playbook, the LLM program asks you the right questions about your business and can implement everything else independently. No prompt engineering, no guesswork.
How complex is the setup?
Setup is quick and easy, as the co-analyst directly accesses your existing minubo data.
If you're already using the minubo suite, you only need to configure the MCP connection in your external LLM (ChatGPT, Claude, etc.). This takes just a few minutes and requires no technical expertise. Our support team is available to assist you if you have any questions. For minubo Chat, it's even simpler: your administrator needs to activate your account once, and then you can get started directly in your minubo app.
How much does the minubo Co-Analyst cost?
The co-analyst is part of the minubo suite – not a separate product. The exact pricing model depends on your individual setup. Contact us for a customized offer.
Can I combine minubo data with other data sources?
Yes, when using external data sources (ChatGPT, Claude, etc.), you can integrate any additional data sources – for example, weather data. The Co-Analyst combines your minubo data with this external information and provides you with a comprehensive analysis.
Important note regarding reliability: While the Co-Analyst delivers the highest quality results for your minubo data (thanks to structured data, semantics, and guardrails), we cannot guarantee this for external data sources. The quality of the comprehensive analysis then depends heavily on the quality and structure of the data you add.
In short: The Co-Analyst can work with anything you provide, but the reliability of the results for external data is your responsibility.
How does the co-analyst prevent the AI from inventing ("hallucinating") numbers?
Through three layers of protection:
1. Structured data: Your minubo data model delivers clean, structured data – no Excel chaos.
2. Semantic model: The AI knows exactly what "contribution margin," "channel," or "return rate" means and how the data is logically structured – no misinterpretations.
3. Guardrails: The LLM receives guidance on how to access your data and correctly apply analytical methodologies. This includes, for example, information on the specific structure of individual data warehouse tables, instructions on how to understand and apply semantic information, general analysis rules, and very specific analytical methodologies for individual business use cases.
In short: The co-analyst doesn't invent numbers because it's trained to work like a real analyst: It understands the question, can classify it within its business context, translate it into correct SQL, and follows a clear methodology.
Can I export results and share them with colleagues?
Yes. You can share analysis results with colleagues as an HTML report, CSV file, or directly as a chat link. Playbook results (e.g., campaign scorecard) are output as a formatted HTML report – ready to share.
What is MCP?
The Model Context Protocol (MCP) is the open technology standard that enables the connection between a Language Management Module (LLM) and your minubo data. It allows language models to securely and seamlessly access server content as well as external data sources and tools via a provided MCP server – in your case, your minubo data warehouse and its associated semantic layer.
What is the minubo Semantic Layer?
The minubo Semantic Layer is a database containing descriptive information about your minubo data warehouse. This includes, for example, overviews of the individual tables or views in the data warehouse, information about each column in the data warehouse, and specific formulas for key performance indicators (KPIs) that are not directly represented in the data warehouse but must be calculated. We have also developed guiding assets for the Semantic Layer that enable a Learning Management Technician (LLM) to manage the large amount of information efficiently and with minimal token usage – while still obtaining all the necessary information for each individual analysis question.
What is the Prompt Coach?
The Prompt Coach is a tool integrated into the MCP (Master's Program in Communication) that empowers the LLM (Linguistic Teacher Training) program to support you in developing analytical prompts – a challenging task, especially for more complex analyses that might also need to be implemented in a reproducible output format. With the Prompt Coach, however, prompt development becomes child's play: It understands all the essential rules of prompt creation, analytical methodology, and potential pitfalls, and is an expert on the minubo data model. This allows it to guide you through the prompt development process, making even complex analyses easily accessible. Best of all, the Prompt Coach adapts to your existing knowledge and can provide support in the most helpful way for you.
ca. 30 minutes
with videocall if desired
