There's a moment in every boardroom that data professionals know well. Someone pulls up a dashboard, the numbers are all there, and then the room goes quiet. Not because the data is bad - but because nobody's quite sure what it's saying.
Charts are brilliant at showing patterns. Tables are brilliant at showing detail. But neither of them can look at a dataset and say: "Here's what actually happened this quarter, here's why it matters, and here's what you should pay attention to next."
That's the gap we set out to close.
The problem: dashboards that show everything but explain nothing
One of our clients - a commercial real estate advisory firm operating across a major US market - had a comprehensive Power BI reporting system. The data was clean, the models were sound, the dashboards were well-built. On paper, everything worked.
In practice, there was a problem.
The people who needed the insights most - brokers, executives, analysts - didn't always have time to sit with a dashboard and interpret what the numbers meant in context. They needed summaries. Written, human-readable summaries that explained what was happening in the market, what had changed since last month, and what deserved attention.
The obvious solution was Microsoft's built-in Smart Narrative feature. There was just one issue: unlocking it required a Fabric capacity licence costing several thousand dollars per month. For a focused advisory firm, that's a significant line item for a single feature - particularly when the built-in version offers limited control over tone, structure, and depth.
So we built something better.
The solution: an AI narrative engine that lives inside Power BI
Rather than paying for a premium licence to access a feature we couldn't fully control, we engineered a custom solution. It sits entirely within the existing Power BI environment, requires no additional Microsoft licensing, and produces narrative summaries that are - quite frankly - more useful than anything the built-in tool delivers.
Here's how it works, in plain terms.
The Power BI report already contains the data - sales volumes, pricing trends, market activity, deal flow, year-on-year comparisons. All the numbers are there, modelled and calculated.
We wrote a layer of logic that takes those numbers - the KPIs, the comparisons, the metrics that matter - and passes them to an AI engine via a secure connection. Along with the data, it sends a set of carefully written instructions: what kind of summary to produce, what tone to use, how long it should be, and what perspective to write from.
The AI processes the data according to those instructions and returns a polished, professional narrative - directly into the Power BI report. The whole thing takes seconds. There's no separate tool to open, no export to run, no manual step whatsoever.
And here's the part that makes it genuinely powerful: when someone changes a filter in the report - selects a different borough, a different quarter, a different property type - the narrative regenerates automatically. New context, new data, new summary. In real time.
Why this matters more than it might sound
The built-in Smart Narrative in Power BI does one thing: it generates a generic text description of whatever's on the page. You can't meaningfully control the language, the depth, or the perspective. It reads like what it is - an automated caption.
Our solution does something fundamentally different.
We can instruct it to write from the perspective of a market analyst, producing a detailed breakdown of trends and anomalies. We can switch that to an executive summary - three paragraphs, no jargon, focused on decisions. We can ask for a broker-facing narrative that highlights opportunities. We can change the language, the length, the focus, and the structure - all by adjusting the instructions in the code.
This isn't a template being filled in. It's a genuine narrative being composed from live data, following precise editorial direction that we define together with the client.
One report. Multiple perspectives. All generated in seconds.
What the client actually gets
Let's be specific about the value, because this is where it becomes difficult to ignore.
Cost. The premium licence required to unlock Microsoft's built-in Smart Narrative feature runs to several thousand dollars per month - upwards of forty to fifty thousand a year. Our solution operates on API usage costs that are, in most cases, under a hundred dollars monthly. That's not a marginal saving. That's an order-of-magnitude difference.
Control. With the built-in feature, you get what Microsoft gives you. With our solution, the client decides exactly how the narrative reads - the tone, the length, the audience, the emphasis. If next quarter they want a different perspective added, we adjust the instructions. No new licence, no platform upgrade, no waiting for a feature request to be acknowledged.
Quality. We've tested both side by side. The built-in feature produces serviceable but generic output. Our narratives read like they were written by someone who understands the business - because, in a meaningful sense, they were. The instructions encode genuine domain knowledge about what matters in commercial real estate, what language professionals in that market use, and what a useful summary actually looks like.
Speed. Filter changes, narrative updates. Seconds. No lag, no loading screens, no "generating your report" waiting states. It feels native to the report because it effectively is.
How we built it - the short version
For those who want the technical outline without the deep dive:
The foundation is a well-architected Power BI data model - multiple data sources connected and transformed through a rigorous ETL process, with a calculation layer that produces the KPIs and metrics the business relies on.
Sitting within that environment is a custom script that extracts the relevant numbers and constructs a structured request - data plus instructions - which it sends to an AI platform via a secure API connection. The response comes back as finished text and renders directly in the report.
The entire interaction is seamless. No external tools, no browser tabs, no copy-paste. The user sees a report with numbers on one side and a professionally written summary on the other - and both update in real time when the context changes.
What this tells you about how we work
We could have recommended the premium licence. It would have been the simpler conversation - "you need this feature, here's what it costs, let's move on." That's what most consultancies would do, because it's easier and it doesn't require building anything.
Instead, we looked at what the client actually needed, evaluated what was available, and concluded that we could deliver something better for a fraction of the cost. Then we built it.
That's not a one-off story. It's how we approach every engagement. The question is never "what's the standard solution?" The question is "what's the right solution for this specific context?" Sometimes the answer is a standard tool. Sometimes it's something that doesn't exist yet.
When it's the latter, we build it.
Is this relevant to your business?
If you're running Power BI reports and wishing they could explain themselves - this is a solved problem. We've built the solution, we've proven it in production, and the economics are compelling.
If you're being quoted thousands per month for a premium licence to unlock a feature that does half of what you need - there may be a better path.
And if you're sitting on a well-built data model and wondering what more it could do for you - the answer is almost certainly "more than you think."
We'd welcome the conversation.