Academy
3. Querying Data in Plain Language
Write natural-language questions against a dataset, read the generated query logic, and interpret the results.
OUPI Data lets you question any imported dataset using everyday language — no SQL or formula syntax required. Open the Query screen and type your question the way you would ask a colleague: "What were total sales by region last quarter?" or "Show me the top 10 customers by revenue." The AI translates your plain-language question into the actual query logic, executes it against your dataset, and returns the results alongside the underlying logic so you can verify exactly what was computed. This transparency means you are never guessing what the system did; you can read the generated logic, confirm it matches your intent, and refine if needed.
Reading the generated query logic is a key skill. After each question, OUPI Data displays the logic it produced — the filters, groupings, aggregations, and sort orders it applied. Review this breakdown before acting on the results: does the date range match what you intended? Is the aggregation a sum when you expected an average? If something looks off, rephrase your question with more precision (e.g., "average order value" instead of just "order value") and run again. Each AI query consumes credits, and the cost is shown before execution, so you always know what you are spending.
Be specific in your questions. Include the metric, the dimension, and any time frame: "Total revenue by product category for Q1 2025" will produce a more accurate query than "Show me revenue." The more context you give, the better the AI translates your intent.
Once you get a result you trust, save the query so you can reuse it later. Saved queries live under the Saved Queries section, where you can re-run them on refreshed data — useful for weekly reports or recurring checks. You can also use a saved query as a starting point: open it, tweak the wording, and run a variant without rebuilding from scratch. This iterative workflow lets you build a personal library of reliable, tested questions over time.
Results are not limited to tables. If a visual would communicate the answer better, ask for one directly in your question — for example, "Show a bar chart of monthly expenses." You can also generate a chart from any result after the fact. Visualizations can be pinned to your main OUPI Data dashboard, turning one-off answers into a living overview you check at a glance.
You can also query your datasets from ChatPro without leaving a conversation. Your datasets are exposed through data tools, so mid-chat you can ask "What was last month's churn rate?" and the AI queries your data hub on the fly.
Go to the Query screen, pick one of your imported datasets, and type a plain-language question (e.g., "Sum of amount by category"). Review the generated logic, check the results, then save the query for future use.
You now know how to: (1) write natural-language questions against a dataset in the Query screen, (2) read the generated query logic to verify filters, aggregations, and sort orders, (3) interpret and trust the results before acting, (4) save queries for reuse and iteration, and (5) optionally turn results into pinned visualizations. Practice refining your questions — specificity is your best tool for accurate answers.