Demo

In this short intro you’ll see how the Power BI Agent chat works. You pick a dataset, type your question in plain language and get an answer right away. The nice part is the form that answer takes: the agent decides for itself what fits best. A clear explanation in text, a tidy table with the figures side by side, and where it helps a chart too — all in the same answer. So you don’t just read what’s going on, you see it immediately. No building visuals, no formulas: you ask a question and the chat presents the result in the clearest way possible.

In this demo you’ll see how Power BI Agent analyzes multiple administrations at once. We ask it to compare the monthly gross margin in 2025 of five hospitality businesses, explain the differences and give advice. On a Fabric dataset the agent first delivers a full monthly table with the margin per company, then a ranking with a profile for each business. It spots patterns, an autumn dip across all locations and two notable outliers, and closes with concrete advice per profile: renegotiate purchasing contracts, analyze costs, apply menu engineering and benchmark against the best-performing location. Cross-administration insight without merging any reports yourself.

In this demo you’ll see how Power BI Agent helps with an urgent liquidity problem. Alongside a cashflow forecast dashboard we ask: from week 21 we’ll run short on cash. What’s the cause and what can we do about it? The agent digs through the forecast and names the causes: a negative opening balance, a large supplier payment, a VAT remittance and purchase orders all landing within a few weeks. It then gives concrete advice, split into short term (bridge financing, spreading payments, collecting receivables faster, deferring VAT) and long term (revising payment terms). From problem to a well-founded action plan in a single question.

In this demo you’ll see how Power BI Agent builds a year-end forecast and tests it against budget. We ask, with a cutoff date of end September 2025, for a full-year 2025 forecast for a single administration. With reasoning on whether the annual budget will be met and recommendations. The agent delivers a clean forecast-versus-budget comparison per line item, flags the pain points (staff costs and car costs over budget) and the upsides (housing and office costs under budget). It gives a clear verdict: the annual budget likely won’t be met, unless Q4 is steered. And it closes with concrete recommendations, each with an impact rating. From loose numbers to a story you can act on.

In this demo you’ll see how a conversation with Power BI Agent deepens. We ask for the development of liquidity and working capital across five wholesale companies over all of 2025. The agent delivers an assessment per company — from strong to concerning — with current, quick and cash ratios and net working capital. One company stands out negatively, so we ask the follow-up question: investigate the causes. The agent analyzes several angles at once — margins, receivable and inventory days, solvency — and concludes the liquidity problem stems from three factors. This way you move from a broad overview to the real root cause, simply by asking on.

In this demo you’ll see how Power BI Agent uses a standard fiscal skill. We ask for the expected corporate tax liability at the end of Q3 2025 for five IT companies. That takes more than adding up numbers. The agent recognizes the fiscal question and uses the built-in corporate tax skill. It applies the current 2025 tax brackets and annualizes the year-to-date result with a seasonal correction — based on last year’s pattern, not a simple straight line. The result is a clean table per company with taxable result, annualized result and expected tax. Fiscal knowledge comes built in, you just ask the question.

In this demo you’ll see the Power BI Agent chat visual, embedded directly in a Power BI dashboard. No separate portal or tab: you ask your question right next to your report. We ask for the key insights into December 2025 revenue. The agent delivers a complete monthly analysis: total revenue versus last year, a top 5 of customers and products, and the revenue split by customer segment. Each with clear tables and a short interpretation of every result. The answer closes with a conclusion and a follow-up question. This gives you a well-founded story behind your numbers in seconds, without building any visuals yourself.

In this demo you’ll see how the Power BI Agent chat visual answers a real “why” question, right inside a sales dashboard. We ask why some sales reps grow faster than others. The agent first retrieves the data — revenue, orders and customers per rep — and then builds a well-founded explanation. Not loose numbers, but an analysis: the difference isn’t about winning new customers, but about order frequency, average order value and performance per customer segment. For each rep the agent names the cause, flags an underperformer that needs attention, and closes with a follow-up question. So you learn not just what is happening, but why.

In this demo you’ll see how to automate a recurring analysis with Workflows. We build a workflow that emails an overview of last week’s top 10 best-selling products every Monday. In the workflow editor you drag nodes onto the canvas and connect them into a logical flow: a scheduled trigger as the starting point, a DAX Query node that retrieves the data, an AI Analysis node that interprets the figures, and an Email node that sends the result. You pass each node’s output along using variables. Set it up once, and from then on the report arrives in your inbox every week.

In this demo you’ll see how a workflow warns you proactively. We build a workflow that surfaces lost customers every month — customers who had no revenue last month but did the month before. The clever part is in the DAX Query node: you don’t have to write DAX, you describe in plain language which data you want and the AI generates the query. The AI Analysis node then interprets the figures, and the Email node sends the overview on. The result: every month an email with the biggest decliners and customers that need immediate follow-up. Spotting churn without having to go looking for it yourself.

In this demo you’ll see how to create a custom skill in Power BI Agent. Skills are analysis modules that tell the agent how to perform a specific calculation — which DAX formulas, which benchmarks and how to interpret the result. Standard skills are detected automatically, but for analyses that are unique to your organization or industry you can build your own. We create the skill “E-bike share of bicycle revenue”, fill in keywords and DAX, and assign it to a dataset. Then we ask a question in the chat about the e-bike trend — the agent uses the skill and immediately returns a well-founded answer with a table and analysis.

In this demo you’ll see how to set up masking in Power BI Agent. Masking lets you decide per field what the AI does and doesn’t get to see — useful for personal data and other sensitive information. You open the masking screen from the dataset menu and choose one of three options per column. Anonymizing replaces the value with *** so the AI can’t read or use the field. Pseudonymizing replaces the value with a consistent pseudonym, so you can still analyze on the field without sharing the real value. And entire tables that aren’t relevant for analysis can be hidden in one click. Masking is applied before data is sent to the AI; your Power BI data stays unchanged.