About AI at Cube – Cube

Cube's AI features are available to all Cube customers and use AI credits from your company's shared annual allotment. See How AI Credits Work

AI features today

The AI Hub is the starting point for AI in the Cube Workspace: ask Charlie a question, run a ready-made FP&A prompt, or hand a modeling task to the Planner.

Note: Charlie is Cube's AI assistant everywhere you use Cube: the Workspace, the Excel and Google Sheets add-ins, Slack, and Teams. On some of these, Charlie may still appear as AI Analyst. It's the same assistant, reading the same governed data and using the same AI credits.

Best practices when using AI tools

When it comes to building and launching AI, we want to be transparent about our approach and how this impacts you.

Here are some important things to keep in mind when using any AI-powered tool:

Use your judgment. Validate all AI-generated suggestions or recommendations before making critical decisions.
Treat output as a starting point. AI-generated analysis should be reviewed by a person, the same way you would review any analysis before it goes to a stakeholder.
Be transparent with your team. Let team members or other stakeholders know when an AI feature has generated content.

How Cube's AI works with your data

Cube's AI features do not guess at your numbers. They read them.

Cube AI reads your real financial data

When you ask a question, Cube AI queries the financial data already in your Cube. It decides which slices of data it needs to answer the question you asked, fetches them, and builds its response from what it retrieves. Answers reflect the actuals, budgets, forecasts, and scenarios in your Cube as they stand at the moment you ask.

Cube AI can also reach transaction-level detail from your connected source systems, so it can move from a summary figure to the underlying line items behind it.

Your numbers stay your numbers

Cube AI reads through the same governed layer as the rest of Cube. That means the figures it works from already have your business logic applied:

Your formulas and calculated members
Your account, department, and entity hierarchies
Your currency conversions, normalized to your reporting currency

It does not recalculate your financials or apply logic of its own. It reads what Cube already holds.

Every answer traces to the source

Because Cube AI reads from your Cube rather than from an external copy of your data, you can follow any figure it returns back through Cube to the source records behind it. Open the report it generates, then filter, drill down, and fetch the detail exactly as you would with any other Cube report.

Access follows your existing permissions

Cube AI is available only to Cube users, and every request runs inside the requesting user's Data Access Scope. It can see what that user can see in Cube, and nothing beyond it. If a user cannot open a scenario, entity, or dimension in Cube, Cube AI cannot read it on their behalf either. Your security roles and dimension-level restrictions apply without any separate configuration.

For more on how access is configured, see security roles in Cube.

Give Cube AI your business context

Cube AI is most effective when it understands the specific nuances of how your organization reports. Admins with the Administer AI permission can set permanent Custom Context through AI Default Preferences, which guides how Cube AI interprets your data and formats its responses. This keeps results aligned with your reporting standards and hierarchy for every user in your organization. The same permission lets admins see company-wide AI credit usage, buy more credits, and choose who receives credit alerts.

See Setting AI Default Preferences.

Review what you get

Grounding responses in your governed data improves them, but it does not make them final. Cube does not warrant the accuracy or completeness of AI-generated output, and every AI surface in Cube carries a reminder to check results before you use them.

Accuracy

Cube's AI features leverage algorithms to identify patterns and relationships from large amounts of data. As powerful as this can be, it's important to remember that the accuracy and reliability of AI-generated results may vary. There is always a possibility of inaccuracies, omissions, or misleading outcomes due to the complexities and limitations of these technologies.

In other words, the information provided by Cube's AI should be regarded as a general reference rather than an absolute fact or a substitute for expert advice or your research.

AI credits

Every AI feature in Cube draws on one shared pool of AI credits for your company. Your contract sets your annual allotment, which resets each year on your subscription start date.

Features that aren't AI, such as fetch, publish, Reports, and Workflows, don't use credits. When your company runs out, AI features pause until credits are added or your allotment resets. Everything else in Cube keeps working.

Track usage on the Usage page in the Workspace side navigation. For details, see How AI Credits Work, View AI Credit Usage, and What Happens When You Run Out of AI Credits.

Who can use Cube AI?

Cube's AI features are available to all Cube customers, and usage draws on your company's AI credits. Individual users may have restricted data and feature permissions based on their organization's security settings. Cube AI permissions are grouped under Cube AI in User Management and include Use Charlie, Use AI Planner, Use MCP (Read Only), Use MCP (Write), and Administer AI. Users without a Cube AI permission can't use AI features and don't use credits. See Manage Cube AI Permissions.

Learn more about security roles in Cube.

Learn more about how we handle data from our trust center.