Video summary
Copilot for Finance: 5 Power Moves You're Not Using
Main summary
Key takeaways
Product Reviewed
Microsoft Excel Copilot — specifically “Excel AI Agent mode” (used inside Excel online)—for:
- Building and auditing financial models
- Consolidating data
- Generating management-ready dashboards
How to Access / Licensing (User Setup Experience)
- You may not see Agent mode immediately after clicking Copilot in Excel online.
- If you see the normal Copilot view:
- Go to Tools
- Activate agent mode
- Microsoft documentation is referenced for which licenses are needed for Excel AI agent mode.
- If Agent mode isn’t available even with the right license:
- Check with your admin (it may need to be enabled/activated).
Key Features Shown
-
Build a revenue model from prompts
- Example prompt: a “CFO-style” prompt to create a subscription revenue model for 3 products:
- Core, Premium, Enterprise
- The agent generates the model using dummy data first, then supports variable changes:
- Price
- Growth
- Churn
- Outputs include ARR, MRR, and growth.
- Example prompt: a “CFO-style” prompt to create a subscription revenue model for 3 products:
-
Dynamic model testing (variable changes propagate)
- Changing variables updates downstream results.
- Example:
- Core growth changing 5% → 30% updates downstream outputs.
- Premium example:
- Updating starting customers (e.g., 500) changes Premium outputs.
-
Audit-first workflow (safe corrections)
- Agent mode can audit the whole model and graphs, identify errors, and suggest improvements.
- The user can instruct it to not fix immediately:
- First: show problems
- Then: fix only if approved
- Fixes happen while avoiding manual formula rewriting.
-
Multi-tab 5-year SaaS 3-statement model generation
- Reuses the revenue tab and creates additional tabs, including:
- Cost per unit
- Head count
- Three financial statements
- Summary tab and graphs
- Headcount assumptions are variable (example includes monthly hire).
- The creator notes formulas are generally not overly complex, keeping the model human-readable.
- Reuses the revenue tab and creates additional tabs, including:
-
Consistency checks via spot-check + AI audit iteration
- Emphasizes reconciling:
- Cash / P&L / balance sheet movements
- Monthly vs yearly movement consistency
- Linked schedules
- Workflow:
- Spot-check first
- Ask Copilot to audit yourself, do not fix yet
- Iterate until acceptable.
- Emphasizes reconciling:
-
Using existing data + consolidation
- The creator argues you can use existing sales data after cleaning/summarizing.
- Agent mode supports:
- Consolidating many tabs into one (“all data”)
- Creating a pivot table summary by month and product (units + revenue/volume)
- Adds an audit tab to verify totals:
- Confirms sums from original tabs match the consolidated data.
-
Management dashboard / presentation outputs
- Prompts generate a new tab with modern graphs for management, such as:
- P&L evolution
- Performance by product
- Sensitivity table (color-coded)
- Example outputs referenced:
- P&L comparisons over time
- EBITDA over time
- Waterfall
- KPIs
- Prompts generate a new tab with modern graphs for management, such as:
Pros (As Stated)
- Time savings
- Auditing tasks that might take humans ~30 minutes (for initial audits) or require weeks of manual consolidation are presented as much faster.
- Reduced risk with audit-first workflow
- Issues are surfaced before changes
- You can request specific fixes only after review
- Variable-driven model
- Changing assumptions updates the model dynamically
- Model stays editable / understandable
- “Easy to understand for a human”
- Formulas are “not too complex”
- Helpful reconciliation and audit documentation
- Creates an auditable structure showing consolidated vs original totals
Cons / Limitations (As Stated)
- Errors can happen
- The creator acknowledges critics may “bash” the tools before verifying results.
- You still must review and iterate
- Spot-checking is required
- Iteration is needed to reach a correct model
- Cash/financial statement consistency must be ensured
- Important caution: save your work / use auto-save
- A “rookie mistake” warning is included—work can be lost if you don’t save.
Comparisons Made
- Versus manual work
- Manual auditing is implied to take ~30 minutes for an initial audit scenario.
- Manual consolidation across very large spreadsheets (e.g., 10 years, 100 tabs) is described as unrealistic.
- Versus “bashing the tools” without testing
- The creator frames AI as iterative and coachable rather than something that replaces responsibility.
- Versus formulas-only approaches
- Pivot table summarization is described as more flexible in this workflow than formula-only summarization.
Unique Points Explicitly Mentioned
- Agent mode may require enabling via Copilot → Tools → activate agent mode.
- Agent mode availability depends on license and possibly admin activation.
- Prompt role context example: “CFO of a software company”.
- Revenue model example includes 3 products and uses dummy data first.
- User can change price, growth, churn.
- Audit flow can be configured to “show problems first, do not fix yet.”
- Example audit issues included:
- Months column values were incorrect
- Chart formatting was wrong (month/year labels didn’t make sense), affecting the chart
- Warning about “competitors/critics” bashing tools before checking.
- Reminders to save data / use auto-save.
- 5-year SaaS model includes additional tabs (cost, headcount, statements, summary, graphs).
- Headcount changes can improve cash-flow realism (e.g., reducing team size).
- Example of unrealistic revenue jump (year 4 → 200M → 1B) attributed to an earlier assumption change.
- Existing data workflow:
- Clean/summarize
- Consolidate to all data
- Generate pivot summary
- Add audit tab to confirm totals match
- Management presentation is emphasized via dashboard-style outputs (PPT-ready graphs).
- Creator concludes: “go try it… you will save so much time.”
Speakers / Multiple Perspectives
- Single main speaker (the creator) throughout.
- The “CFO” framing appears as a prompt role, not as a separate human speaker.
Overall Verdict / Recommendation
Recommended for finance workflows in Excel—especially when you need to generate, audit, and iterate financial models quickly.
The video emphasizes that while errors can occur, the audit-first prompt approach, spot-checking, and iteration make Excel AI Agent mode significantly faster and more reliable than doing everything manually.