What the Mizan Healthcare AI Analyst can answer, how to ask good questions, and what its limits are.
The Mizan Healthcare AI Analyst is a deterministic natural language interface grounded in the local Mizan Healthcare demo model. You can ask it questions in plain English or Arabic and get structured answers in tables, summaries, and trend explanations without navigating dashboards manually.
Click 'Ask Me' in the left sidebar, or navigate to Profitability > Ask Me. You can also open the chat widget from the header bar on any page. The assistant has context about your current filters, including date range, Mizan Healthcare entity, and department.
Good questions: 'Which Mizan Healthcare segment has the best margin?', 'What is driving Mizan Healthcare facility cost variance?', 'Show me revenue by department for the current year.' The assistant retrieves synthetic Mizan Healthcare demo numbers from the local source model and presents them in a structured table or summary.
Responses include the data table or summary, a brief interpretation, and sometimes a recommended action. If the AI is uncertain about the interpretation, it will say so. Always verify critical numbers against the dashboard directly before using them in external reports.
The AI assistant cannot access external data, make future predictions, or take actions like generating reports or changing settings. It only answers questions covered by the synthetic Mizan Healthcare demo model. Questions about audited actuals, live market data, or external sources are handled as out of scope.
Be specific about the time period: 'last quarter' is better than 'recently'. Use month names or quarters for clarity.
If you get an unexpected answer, try rephrasing with the exact metric name as shown in the dashboard (e.g., 'Net Income' instead of 'profit').
The AI supports Arabic — you can ask questions in Arabic and receive answers in Arabic, including Arabic numerals and Saudi Riyal currency formatting.
For complex multi-step analysis, break it into separate questions rather than asking everything at once.