AI portfolio data
AI Made Portfolio Analytics Easy.
Trusting the Answer Is the Harder Part.
Connecting an AI to your live data made getting an answer easy. It didn’t make the answer trustworthy. Those are two different problems.
You might be wondering how an AI chat can even answer questions using your AI portfolio data. Your data isn’t sitting out on the open web for any model to find. It can, because we connect the AI directly to where your numbers actually live: your property management system. Open a chat window, type your question in plain English, and an answer comes back in seconds. That part is solved.
There’s another question that’s harder to answer: can you trust that number enough to act on it?
Speed and trust. Connecting an AI to your live data made getting an answer easy. It didn’t make the answer trustworthy. Those are two different problems, and most teams are only solving the first one.
Here are the two ways we’ll have you converse with AI about your portfolio, and when to use each one.
RentViewer gives you both lanes, on purpose.
Explore
A live, direct connection to your property management system for fast answers. Ask a question, get a number, move on. No setup, no waiting on a report, no analyst in the loop. This is the fastest way to look around your portfolio, test an idea, or answer a one-off question from a board member.
Operationalize
A managed data foundation that powers anything that becomes recurring, investor-facing, or consequential. Once a number leaves your desk and starts showing up in a board deck, a lender covenant, or a monthly KPI review, it needs a definition that doesn’t shift depending on who asked, a history that reconciles, and a record of where it came from. If you have worked with us before, you know about our “Metrics Specification” document. That’s what we are talking about here, along with change control, audit trails, and evidence that the numbers were verified. That tight data management process is being called “governance” in the press, so we have also adopted that term.
Neither lane is the “real” one. Explore is not a lesser cousin of Operationalize: it’s the front door. The point is to make exploration effortless, then give you the infrastructure to promote what you find into something you can stand behind.
This holds whether you’re running multifamily asset management or tracking manufactured housing data; the line between explore and operationalize looks the same either way.
Explorer, Steward, Scaler
We talk about three needs a person brings to portfolio data: Explorer (move fast, ask anything), Steward (the numbers have to be right and defensible), and Scaler (build something that holds up over time). These aren’t job titles. They’re modes. A CEO chasing a hunch on a Sunday night is in Explorer mode. That same CEO presenting to a lender on Monday is in Steward mode. And the same CEO concerned about hundreds of chats and generated outputs cluttering everyone’s laptop is in Scaler mode. Your analysts might carry all three needs in a single afternoon.
RentViewer is built around the needs, not the org chart. That’s why Explore and Operationalize sit side by side instead of one gating the other.
Why “access” quietly becomes a risk
A person can steer an AI model until it agrees with them.
We tested our own live connection to Rent Manager against more than 30 controlled test cases, and it performed accurately across all of them. But those cases were deliberately built to be unambiguous: clear questions with one right answer. Real use is messier. In beta, customers sometimes wrote a prompt that was open to more than one reading. When the answer came back different from what they expected, some rewrote the question until the answer matched what they’d assumed going in.
That’s not a knock on the customer, and it’s not a flaw in the model. It’s a completely human instinct; we all do it when a number surprises us. But it means an unstructured connection to live data can quietly produce a confirmation-bias problem instead of an accuracy problem. The AI didn’t get it wrong. The question got rewritten until it agreed with the person asking it.
That’s exactly the scenario governance exists for. Not to slow down exploration; to catch the moment a number is about to become consequential, and make sure it’s been asked once instead of asked until it felt right.
What this looks like in practice
You don’t need governance on every question. You need it on the questions that matter. A few examples of where the line usually falls:
- Which unit statuses are included in occupancy calculations
- Which GL accounts comprise CapEx
- Which properties belong in a cohort of their own and can’t be compared to other properties
- Whether the rent amount collected should come from the GL or from the Rent Roll
- How a move-in reconciles against a signed lease
And so on. None of these are edge cases; they’re the everyday judgment calls that make a KPI mean the same thing every time someone asks for it.
Explore first. Ask anything, as often as you want. When a number is about to show up in a report, a recurring review, or a decision: that’s when it moves into the governed lane.
Frequently asked
Chat with your portfolio data, freely.
One trusted source of truth for your portfolio: investor-ready reports, production dashboards, and AI chat your team can rely on, without hiring a data team. That’s the whole idea in one line. Build what matters on a foundation you can stand behind.