oneAgent combines understanding with reliability
This is how AI works with business intelligence.
oneAgent understands questions in natural language and answers with the metrics you have verified and approved. It delivers your numbers unaltered all the way to the business teams and helps them turn the numbers into targets and deviations into tasks.
- Hosted in Germany or on your own servers, including the language model
- Built on more than 150 BI projects over the past five years
- Answers in German and English
How does this work in practice?
The AI understands the question. The answer is the verified metric.
One definition, approved once, applies to every answer, every follow-up question, every target and every task.
Approved
The BI team defines the metric and approves the definition. Only then does it carry the “Verified metric” seal.
View exampleHide example
Asked
Anyone in the company who needs a number asks in their own words. If the question is ambiguous, oneAgent asks back.
View exampleHide example
Analyzed
oneAgent suggests what to look at next until it becomes clear where a deviation lies.
View exampleHide example
Followed up
A target can be set for any metric. If the metric deviates from it, oneAgent flags it, drafts a task for the people responsible and follows up until it is done.
View exampleHide example
What gets better for you and for your company?
Your definition applies to every question and creates clarity across the entire organization.
From the hundredth report to root-cause analysis
Your metrics reach the people who need them. This makes side calculations in Excel or ChatGPT unnecessary. You can focus on your actual work and rely on oneAgent to answer with your definition.
How to set up a metricHow your colleagues benefit from oneAgent
Experience how oneAgent would work with your metrics.
Book a demoWho is behind oneAgent?

oneAgent grew out of our work in BI projects.
At first, we tried to keep improving the AI’s answers. But with numbers, almost right isn’t good enough. That experience led to what you have seen above.
How does oneAgent fit into your IT landscape?
oneAgent connects to your existing data warehouse and source systems, including several at once. It runs on servers in Germany or on your own, language model included.
Cloud servers in Germany
Data and language model are hosted in data centers in Germany.
Your own servers
Everything stays in your network, including the language model.
oneAgent uses them to connect ERP, CRM, shops, data warehouses, databases, files and APIs, either live at the source system or via import into the oneAgent Lake.
- Microsoft Fabric
- SQL Server
- Snowflake
- Databricks
- PostgreSQL
- Oracle
- Salesforce
- HubSpot
- Shopify
Frequently asked questions about oneAgent
What is oneAgent?
oneAgent is an AI assistant for business intelligence from oneLake GmbH. Employees ask questions in natural language via chat and get answers from metrics that responsible teams such as BI or controlling have defined, verified and approved beforehand. Unlike generic AI chatbots, oneAgent separates language understanding from the metrics: the AI understands the question, and the number comes from the verified definition. oneAgent also keeps track of targets, flags deviations and follows up on tasks. It connects to more than 550 data sources, including ERP, CRM, shops and data warehouses, and runs on cloud servers in Germany or on your own servers.
Who is oneAgent for?
oneAgent is designed for mid-sized companies up to the upper mid-market that define and approve their metrics, usually based on a data warehouse or BI system. It serves two groups: business teams and management, who need fast and reliable answers, and BI, controlling and data teams, who are responsible for metrics, permissions and governance. oneAgent is available in German and English.
Can oneAgent hallucinate metrics?
Not with verified metrics. oneAgent uses AI to understand questions. The number itself comes from the approved definition with its source, business rules and filters, and it carries the “Verified metric” seal. If a question is ambiguous, oneAgent asks back instead of guessing. Analyses outside verified metrics are only available on explicit request, and oneAgent labels them as not verified.
How is your own data connected?
There are two ways: live at the source system or via import into the oneAgent Lake. With a live connection, oneAgent queries the source system, such as your data warehouse, directly without moving the data into a second system. With an import, the data is loaded into the oneAgent Lake through an ETL process and prepared there for querying. Which option makes sense depends on the data source, how current the data needs to be, performance and security requirements. Several sources can be connected at the same time.
How is oneAgent introduced, and how much effort is involved?
The introduction starts with connecting a first data source. The business team, controlling or BI then define the source, calculation, filters and owners for each metric and approve it. From that moment on, oneAgent returns the same value for that metric in every question. You don’t start from scratch, because your existing data model is the foundation. Nor do you have to map your entire set of metrics: getting started usually means focusing on the few metrics that are actually discussed in your management meetings. Your own data is connected jointly as part of a pilot project.
Does oneAgent replace our dashboards?
No. Your data warehouse, BI tools and dashboards remain the backbone. oneAgent sits on top as an AI layer and answers the questions that fall between reports, using the same verified metrics. You don’t have to rebuild your data landscape.
What does oneAgent cost?
The cost depends mainly on the deployment model and the number of users. The current guide price is €25 per user per month in the cloud. On your own servers and with larger user numbers, the price can be lower. In the demo, you receive an offer that fits your starting point.
Where is the data processed?
You decide. oneAgent runs on cloud servers in Germany or entirely on your own servers, including the language model in both cases. With a live connection, your data stays in your source system. A role and permission concept controls who can see which metrics, and there are no uncontrolled exports to external AI tools.
How do we get started?
Tell us the metric that sparks the most debate in your company. In the demo, you will experience how it can become a verified metric that nobody needs to argue about anymore.
