Revan Analytics by DataMaaS
Questions people ask
The short answers. Where a question has a longer answer elsewhere, there’s a link.
What it is
What Revan is, and what it isn’t.
What is Revan?
An analyst and a strategist in one system. The analyst reads your data, establishes which basis the question needs, and ties every figure to where it came from. The strategist tells you what it means and what to do about it. You ask a question in plain English and get back a recommendation you can defend, not a table you still have to interpret.
How is it different from ChatGPT, Copilot, or Gemini?
Revan runs on frontier AI, so the difference isn’t the engine. It’s the operating system wrapped around it: your governed data instead of general knowledge, your KPI definitions instead of a guess from a column name, evidence rules that stop it inventing figures, and ten fixed lenses every answer is stress-tested from before it reaches you. A general AI assistant will hand you a confident number it made up. Revan is built so it can’t.
So why not just use a general AI assistant directly?
Because the model is the easy part. A frontier model on its own doesn’t know your schema, doesn’t know that your business defines margin differently from the way the column name suggests, and has no access to your systems — so you’d be pasting figures in by hand, which is exactly where the errors come from.
It also has no reason to be careful. Ask any general AI assistant for a number and it will give you one, because being helpful is what it optimizes for. Revan is built the other way: it will tell you the data can’t answer your question, which basis a figure is reported on when there is more than one, that a figure is an estimate rather than a measurement, and that your premise doesn’t hold. That behavior isn’t in the model — it’s in the rules around it.
And it’s under your entitlements, every time. Same engine, a different standard of answer.
Does it replace my BI tool or my analyst?
Neither. Your BI tool shows you what happened; Revan tells you what it means and what to do next. Your analyst still owns the judgment call — Revan gets them to the defensible version faster, and hands them the evidence trail when someone challenges the number.
Trust
Whether you can rely on the answer.
What if Revan is wrong?
You’ll usually know before it matters, because every material answer carries what it can’t yet prove, how confident it is, and the one variable most likely to change the conclusion. Ten fixed lenses are applied to every answer before it reaches you, and where one of them materially disagrees, the caveat arrives with the conclusion. When Revan is shown to be wrong it corrects, says so plainly, and updates anything that depended on it.
How Revan thinks →
How do I know it isn’t making numbers up?
Because it won’t state a figure no source supports. If the data doesn’t contain what you asked for, Revan says so and names what would be needed — it doesn’t fill the gap to complete the story. Every material figure arrives with its source and basis attached, so you can check it rather than trust it.
Will it tell me if I’m wrong?
Yes, and this is deliberate. If your premise doesn’t hold, you’ll hear it, with the evidence attached. Agreeable analysis is worth nothing — you can already get that anywhere.
What if my question isn’t clear enough?
Revan asks before it answers. Where something decision-critical is missing — which period, which basis, whose numbers — it will ask a short question rather than guessing and presenting the guess as fact.
Data & security
What it can see, and where things live.
What can Revan see?
Only what your role permits. Entitlements apply at three levels — which tables exist for you, which rows you can see, and which columns are gated — and they’re enforced at the database before any result returns. Excluded data doesn’t exist for you, and Revan can’t tell that anything was filtered out.
How your data is read →
Is my data used to train AI?
No. Revan reaches the model through the AI provider’s API, under commercial terms that contractually exclude your prompts and outputs from model training. It isn’t a setting you switch on, and it isn’t something that can be changed for you.
How your data is read →
Where is our data held?
Each Revan organization defines where its data sits. The default is the United States. If you run Revan on your own infrastructure, your data stays wherever that infrastructure is.
Can you run Revan on-premise?
Yes, with an enterprise license. It adds setup and ongoing maintenance, so it’s worth it when data residency or sovereignty requires it rather than as a default. Most customers are better served by the hosted platform — talk to DataMaaS about which fits.
Projects
Bringing your own data.
Your own data, without waiting for anything to be connected.
What is a project?
A place to bring your own data and work on it immediately. Create a project, upload a file or several, and ask questions against them straight away.
How does Revan know what my data means?
You tell it. Each project takes instructions where you describe the data in plain English: what a field actually represents, how your business defines a metric, what to include and what to leave out, which basis a figure is reported on. That’s the same job DataMaaS does when it builds a governed data layer for you — done by you, in your own words.
Can I share a project with my team?
Yes. Share it and colleagues can run their own analysis against the same files and the same instructions, so everyone is working from one definition rather than four spreadsheets that have quietly diverged.
Who can see a shared project?
Everyone you share it with. This is the one place to be careful: a project is governed by what you put in it and who you share it with, not by the role-based entitlements that govern a governed data layer. If a file shouldn’t be seen company-wide, don’t put it in a project that is.
When would I use a governed data layer instead?
When the data lives in your systems rather than in a file, when it needs to stay current, or when different people should see different slices of it. A governed data layer refreshes on a schedule you set — usually daily — so you’re never answering from a snapshot someone uploaded three weeks ago. Projects are for moving fast on data you already have in hand; a governed data layer is for making your live systems answerable.
Getting started
Starting, and what happens after.
What do I need to have in place?
A sign-up and a question worth asking. You don’t need a data project first: you can work on our demo data — a manufacturer’s ERP or a managed service provider’s PSA — from the first hour, or create your own project and upload files. Connecting your own systems is separate work and isn’t required to get value out of the trial.
What does it cost?
Seven days free, then Revan Analytics starts at $249 per month for your company: one subscription, shared by your users. Billing follows the calendar month, so a mid-month start is prorated. Building a governed data layer from your own systems is a separate, scoped engagement.
What happens after the 7 days?
Your subscription starts, unless you have cancelled. Once your account is provisioned we send you a link to add a payment card, and nothing is charged during the 7 days. To cancel, email revan@datamaas.com before they end. Cancelling closes the account, and closing it removes the account and all of its data and history. If you’re already a DataMaaS customer, speak to your usual contact; there may be nothing to arrange.
Where next
Keep going.
Data & security
How your data is read →
How a question reaches your data, what comes back, and who decides — read-only, under the permissions you already set.
Judgment
How Revan thinks →
The ten reflex lenses — the angles every answer is stress-tested from before it reaches you, and how ten views resolve into one recommendation.
Still have a question?
Bring it to Revan directly — or email revan@datamaas.com and a person will answer.
