You are viewing a dashboard and want to understand which products explain a variance, which customers are growing or how to summarise results for management. With EP AI Analyst, the AI extension for EP-Reporter, the question starts in the same Power BI or Qlik Sense report you use every day.
Dashboard context becomes part of the conversation
EP AI Analyst uses data from the current view as the basis for the conversation: filters, selections, measures and rows available in the visual. The configured context is transferred automatically, so users can focus on their question without describing what they are looking at each time.
Four ways to make your report more useful
- Understand: ask which areas, customers or products are performing best and get a concise overview.
- Explain: analyse anomalies and variances between Actual, Budget and Forecast.
- Summarise: generate management commentary and summaries in the user’s chosen language.
- Explore: continue the conversation with follow-up questions in the same analytical context.
How it works: from visual to answer
The question starts in the report
The user interacts with a Power BI Custom Visual or a Qlik Sense extension.
The component gathers the context
Visible data, filters, selections and metadata form the configured scope.
E-Project services apply the rules
APIs and orchestration manage prompts, settings and the AI model chosen by the customer.
The conversation continues
The user receives the answer and can explore the results while retaining the analytical context.
The visual and extension remain lightweight: configurations, provider credentials and usage controls stay in backend services. The solution integrates with the existing BI environment.
The customer chooses the model and governs usage
The console lets customers configure assistants and use cases, select a provider and model, customise system prompts, define permitted reports and fields, and set rules and limits. Available providers include Azure OpenAI, OpenAI, Claude and Gemini.
Request history, tokens consumed per request and usage analysis help track adoption and consumption according to customer settings.
Managed AI or your corporate AI service
There are two adoption options: Managed AI, to start with a managed configuration, and BYO AI, to connect your own corporate provider. The choice depends on the agreed plan, data policies and consumption limits defined for the project.
Security: an explicit scope before launch
Implementation defines which data is transferred, which provider processes it and which information is logged. LLM provider credentials remain in the backend; a manageable client key identifies the integration.
Onboarding checks access, network connectivity, reports and permitted fields. Logging, retention and deployment are defined during the security assessment. Validating answers with the customer’s team is part of the launch process.
An adoption journey in six steps
- Use case: reports, users, questions and expected outcomes.
- Access: tenant, credentials and a dedicated API key.
- Configuration: LLM model, prompts, rules and data scope.
- Integration: installation of the visual or extension and connection to E-Project services.
- Validation: testing answers, consumption and controls.
- Go-live: monitoring, support and improvement.
E-Project handles technical configuration, integration and testing with the customer’s team. The customer provides the report and a business contact, defines policies and the provider, and validates answers before accepting the release.