Ready-to-use prompts for Claude and ChatGPT that connect to your Cube data — from quick analysis to board-ready deliverables.
Cube's integrations with Claude (Anthropic) and ChatGPT (OpenAI) let you query your financial data using your favorite AI tools. You can ask an AI assistant to pull actuals, compare scenarios, and generate deliverables — all powered by your live Cube data through MCP (Model Context Protocol).
This article provides example prompts across common FP&A use cases to help you get productive fast. Each prompt is designed to work with any Cube company — just connect your account and start asking.
💡 Prerequisites: Make sure your Cube MCP connection is set up before trying these prompts. For setup instructions, see Connect Cube to AI Apps via MCP Server.
The write-back prompts below also require the Use MCP (Write) permission.
Claude + Cube: Top Prompts
Claude excels at combining Cube data pulls with structured analysis, file creation, and polished deliverables. These prompts showcase what's possible when you pair Claude's reasoning with your live financial data.
Build a Board Deck from Live Data
BOARD DECK
Prompt: "Build me a board deck showing this quarter's performance. Include revenue by quarter, gross margin trends, opex breakdown by department, and headcount growth. Compare actuals to budget."
Why it works: Claude pulls multiple dimensions from Cube, structures the narrative, and generates a downloadable PPTX — turning what used to be hours of manual work into a single conversation.
Actuals vs. Budget Variance Analysis
FORECASTING & ANALYSIS
Prompt: "Compare our actuals to budget and forecast for this year. Show me the variance by quarter for Revenue, COGS, and Operating Expenses — which line items are most over/under?"
Why it works: This is the bread-and-butter FP&A question. Claude will pull both scenarios from Cube, compute the variances, and highlight the biggest deltas so you know exactly where to focus.
Model a Cost Reduction
WHAT-IF SCENARIO
Prompt: "If we cut Sales & Marketing spend by 15% in the back half of the year, how would that change our Net Income? Compare to what we actually spent."
Why it works: Claude fetches your actuals, applies the adjustment, and shows the downstream P&L impact — ideal for pressure-testing decisions before your next planning cycle.
Interactive KPI Dashboard
DASHBOARD
Prompt: "Build me an interactive dashboard showing monthly revenue trend, operating expense categories as a percentage of revenue, and gross margin % by quarter."
Why it works: Claude creates a live, interactive dashboard directly in the conversation. You can interact with the charts, then iterate on the design by asking for changes in plain English.
ChatGPT + Cube: Top Prompts
ChatGPT brings its own set of strengths to Cube — particularly Deep Research for external benchmarking, Canvas for iterative document editing, GPT-5's extended thinking for complex scenario modeling, and code execution for ad-hoc data analysis.
Internal Data Meets External Benchmarks
DEEP RESEARCH + CUBE
Prompt: "Use deep research to analyze our company's financial performance in the context of current market conditions. Pull our revenue, margins, and opex from Cube, then research comparable SaaS benchmarks and write a board-ready market positioning report."
Why it works: Deep Research autonomously browses hundreds of web sources for 5–30 minutes, then synthesizes findings with your live Cube financials. The result is an analyst-grade report that blends internal data with external benchmarks.
Collaborative Variance Report
CANVAS + DATA ANALYSIS
Prompt: "Pull our quarterly P&L from Cube for 2024 and 2025 and open it in Canvas. Build a variance analysis with charts showing actuals vs budget by department, and let me edit the commentary inline before exporting to PDF."
Why it works: Canvas gives you a side-by-side editing workspace. ChatGPT pulls the data and drafts the analysis, then you refine the narrative, tweak charts, and export to PDF or DOCX — all without leaving the conversation.
Multi-Scenario Headcount Modeling
GPT-5 THINKING + SCENARIO MODELING
Prompt: "Connect to Cube and pull our current headcount, personnel costs, and revenue by department. Then model three scenarios: (1) 10% RIF in Sales, (2) 20% Engineering hiring ramp, (3) flat headcount with 5% raises. Show the net income impact of each."
Why it works: GPT-5's Thinking mode applies multi-step reasoning to chain data pulls, build scenario math, and present structured trade-off analysis — ideal for workforce planning sessions.
Trend Analysis with Charts
CODE EXECUTION + VISUALIZATION
Prompt: "Pull monthly revenue by product line and market from Cube for the last 12 months. Run a Python analysis to identify seasonal patterns, calculate month-over-month growth rates, and generate an interactive chart with trendlines for each product."
Why it works: ChatGPT writes and executes Python (pandas, matplotlib) directly on your live Cube data — producing interactive charts and statistical analysis in a single turn, no coding required.
Write-back Prompts
These prompts change data in Cube and require the Use MCP (Write) permission. A write lands only in a scenario that is open to AI write-back, or one the AI creates. If a write is blocked because the target scenario is not open to AI, the AI will tell you the fix: open the scenario in Cube (set its write protection to "AI"), or have the AI create a new scenario. For how this works, see Connect Cube to AI Apps via MCP Server.
Reforecast Write-back
WRITE-BACK
Prompt: "Pull the latest actuals through last month, apply my revised assumptions (attached), and write the updated Q3 reforecast into our FY26 Forecast scenario."
Why it works: The AI updates the forecast scenario in place, so you skip the manual re-entry. The changes appear in your audit trail attributed to you and flagged as AI-originated. FY26 Forecast must be open to AI write-back first.
Build a Downside Scenario from Scratch
SCENARIO CREATION
Prompt: "Create a new scenario called Headcount Downside, then write in a plan that assumes a 10% reduction in Sales headcount and a hiring freeze in Engineering. Show me the rolled-up impact on operating income."
Why it works: Because the AI creates the scenario, it is opened for AI write-back automatically, so the AI can populate it with no manual setup, then read back the impact for you to review.
Create a Tag for Reporting
TAGGING
Prompt: "Create a tag called Strategic Accounts that groups our top ten revenue accounts, so I can filter reports by it."
Why it works: The AI creates a new grouping you can use across reports. Tags sit outside the account hierarchy, so creating one is additive and does not change any rolled-up total.
Build a Board in Cube
BOARD BUILDING
Prompt: "Build a Board in Cube showing Revenue, Gross Margin, and Operating Expenses by quarter for this year versus budget. Use a combo chart for revenue and a table for the variance."
Why it works: Unlike a static export, this produces a real, live Board in your Cube workspace that you can open, share, and edit like any Board you built yourself.
Tips for Writing Great Prompts
These prompts are ready to use as-is, but you'll get even better results when you tailor them to your business:
| Tip | Example |
|---|---|
| Be specific about time periods | "Q2 2025" or "the last 12 months" rather than "recently" |
| Name your scenarios | "Compare Actuals to Budget v2" instead of "compare to plan" |
| Specify the output format | "Build a slide deck" vs. "show me a dashboard" vs. "give me a table" |
| Layer on follow-ups | Start broad ("show me revenue by quarter"), then drill in ("now break that down by entity and product") |
| Reference your dimensions | Use your Cube dimension names (departments, entities, products, markets) for more precise results |
| Name the target when writing back | "Write this into FY26 Forecast" so the AI updates the scenario you intend |
What's Next?
These prompts are just a starting point. As you get comfortable, try combining multiple asks in a single prompt — for example, pulling data, analyzing it, and generating a deliverable all at once. Both Claude and ChatGPT will handle the multi-step workflow for you.
If you have a use case that isn't covered here or need help writing a prompt for your specific workflow, reach out to your Cube account team — we're happy to help.