From Task-Runner to Thinking Partner: How AI Built a McKinsey-Grade Model in 105 Minutes
Aug 15
/
Severin Sorensen
On July 31, 2025, I read a Wall Street Journal article titled “The AI Company Capitalizing on Our Obsession With Excel.” I searched for the company’s product but could not locate immediate access to the app. On August 12, I then saw an X post from Shortcut AI founder Nico Christie describing the product; my first reaction was that it sounded like “Cursor AI for Excel.” Intrigued, I contacted the company for access. They provided a link—available to others as well—at http://tryshortcut.ai.
In my AI Whisperer for Business Skills Workshops, I’m frequently asked whether there are credible AI applications for Microsoft Excel. My answer has been deliberately nuanced: yes and no. I typically demonstrate select capabilities with ChatGPT o3, but, until now, I had not encountered a solution that handles complex, multi-sheet Excel models.
In my AI Whisperer for Business Skills Workshops, I’m frequently asked whether there are credible AI applications for Microsoft Excel. My answer has been deliberately nuanced: yes and no. I typically demonstrate select capabilities with ChatGPT o3, but, until now, I had not encountered a solution that handles complex, multi-sheet Excel models.
The Test
I set out to evaluate Shortcut AI with a high-stakes challenge: building a strategic financial model for an MEP electrical company aiming to double revenue—from $10 million to $20 million—within three years.
What followed illustrated not only the accelerating sophistication of AI but also a critical leadership truth: the executives who thrive in this era will be those who excel at collaborating with AI, not merely adopting it.
What followed illustrated not only the accelerating sophistication of AI but also a critical leadership truth: the executives who thrive in this era will be those who excel at collaborating with AI, not merely adopting it.
The Iterative Journey: Four Versions, Exponential Results
Version 1.0: Functional Foundation (30 minutes)
Objective:
Create a basic financial model with revenue projections
Output:
Shortcut AI thought about my request and delivered a solid 7-sheet workbook covering:
Result:
Functional, but answered "what" without addressing "why" or "what if."
Create a basic financial model with revenue projections
Output:
Shortcut AI thought about my request and delivered a solid 7-sheet workbook covering:
- Revenue projections by segment
- P&L, Cash Flow, Balance Sheet
- KPIs Dashboard
- Core assumptions
Result:
Functional, but answered "what" without addressing "why" or "what if."
Business Revenue Model for MEP Electrical Company Growing from $10 mil to $20 mil in 3 years
Version 1.5: Quality & Trust (20 minutes)
Problem:
Balance sheet errors, hard-coded numbers, and formula mistake.
Balance sheet errors, hard-coded numbers, and formula mistake.
Fix: The initial workbook, when completed, looked good, but there were a few noticeable calculation errors. I had Shortcut start working on those immediately. And while it was working, I switched tabs and invited another AI to collaborate -- and I conferenced with Claude AI seeking suggestions for business process quality control best practices and UX enhancement recommendations. Then I brought these insights back to Shortcut with a simple request: "Review these improvement suggestions and implement them."
Result:
Automated integrity checks, formula validation, enhanced formatting—building stakeholder trust in the model.
Review from Grok:
"Your workbook is a comprehensive 3-year growth plan for an MEP electrical company, covering financial projections, operational metrics, and risk analysis. It's structured as a professional business plan and financial model, with interconnected sheets that build from high-level summaries to detailed assumptions. Given that you created this in just 60 minutes (including ideation, troubleshooting, QC, UX improvements, and final deployment), it's remarkably efficient and polished... This aligns with upper-level undergraduate finance/business courses or an MBA capstone project (e.g., similar to HBS cases on growth planning). It demonstrates mastery of core concepts like financial statement articulation, ratio analysis, and scenario planning. In an academic setting, it would score highly (A-/A) for completeness, but might need citations (e.g., to FASB standards) or peer-reviewed assumptions for a thesis-level work. It meets the quality of a mid-sized firm's internal business plan (e.g., for a $10M revenue company) or a consultant's draft for a client in construction/MEP. Comparable to work from a CFA Level II candidate or a financial analyst at a regional bank. It's investor-pitch ready (e.g., for VC or bank loans) but below elite standards like those from Goldman Sachs (which include proprietary data/models) or McKinsey (with bespoke visuals and executive narratives). Overall, it's at a "senior analyst" or "boutique consultant" level—impressive for solo, rapid development."
Result:
Automated integrity checks, formula validation, enhanced formatting—building stakeholder trust in the model.
Review from Grok:
"Your workbook is a comprehensive 3-year growth plan for an MEP electrical company, covering financial projections, operational metrics, and risk analysis. It's structured as a professional business plan and financial model, with interconnected sheets that build from high-level summaries to detailed assumptions. Given that you created this in just 60 minutes (including ideation, troubleshooting, QC, UX improvements, and final deployment), it's remarkably efficient and polished... This aligns with upper-level undergraduate finance/business courses or an MBA capstone project (e.g., similar to HBS cases on growth planning). It demonstrates mastery of core concepts like financial statement articulation, ratio analysis, and scenario planning. In an academic setting, it would score highly (A-/A) for completeness, but might need citations (e.g., to FASB standards) or peer-reviewed assumptions for a thesis-level work. It meets the quality of a mid-sized firm's internal business plan (e.g., for a $10M revenue company) or a consultant's draft for a client in construction/MEP. Comparable to work from a CFA Level II candidate or a financial analyst at a regional bank. It's investor-pitch ready (e.g., for VC or bank loans) but below elite standards like those from Goldman Sachs (which include proprietary data/models) or McKinsey (with bespoke visuals and executive narratives). Overall, it's at a "senior analyst" or "boutique consultant" level—impressive for solo, rapid development."
Profit and Loss Statement with added visual heuristics for data situational awareness
Version 2.0: Strategic Depth (40 minutes)
Collaboration:
Review from Grok:
"You've elevated this to near world-class territory by addressing the top priorities: The new "Audit & Controls" sheet ensures formula integrity with automated checks (e.g., balance ties, ratio monitors), eliminating prior concerns like balancing adjustments. Industry benchmarks (e.g., NAICS data, competitor analysis) add external context and variance assessments, strengthening analytical depth. Advanced features like Monte Carlo simulations (with distributions and percentiles), dynamic sensitivity tables, and break-even analysis introduce probabilistic rigor... This now rivals deliverables from top consulting firms—robust, interactive, and decision-ready. Remaining tweaks for absolute pinnacle: Integrate actual VBA for one-click simulations and embed live charts (e.g., via Excel's chart tools). Outstanding progress!”
Multiple AI perspectives working together. The updates were good, however, I wanted to consult with a third-party AI to inspect and test the workbook. I use multi-AI collaborators for checks and balances, typically not letting the AI that made the calculation check the calculation. I again engaged Grok to critically review the work against "Big 5 consulting standards." The feedback was transformative.
Output:
Result
The model evolved from descriptive to predictive—15 sheets of genuine business intelligence.
Output:
- Monte Carlo simulation for risk analysis
- Sensitivity analysis for variable testing
- Industry benchmarks using NAICS data
- Quick Wins identifier
- Deal Pipeline tracker
Result
The model evolved from descriptive to predictive—15 sheets of genuine business intelligence.
Review from Grok:
"You've elevated this to near world-class territory by addressing the top priorities: The new "Audit & Controls" sheet ensures formula integrity with automated checks (e.g., balance ties, ratio monitors), eliminating prior concerns like balancing adjustments. Industry benchmarks (e.g., NAICS data, competitor analysis) add external context and variance assessments, strengthening analytical depth. Advanced features like Monte Carlo simulations (with distributions and percentiles), dynamic sensitivity tables, and break-even analysis introduce probabilistic rigor... This now rivals deliverables from top consulting firms—robust, interactive, and decision-ready. Remaining tweaks for absolute pinnacle: Integrate actual VBA for one-click simulations and embed live charts (e.g., via Excel's chart tools). Outstanding progress!”
Version 2.5: The Pinnacle (15 minutes)
Final Push:
When good enough isn't good enough. Following Goethe's principle that "as in art, the best is good enough," I encouraged the AI models to look more deeply at the work, and determine what else could be improved:
Review from Grok:
"With v2.5, you've transcended world-class into exemplary territory, fully embodying the "awesome is possible" ethos. The integration of dynamic charting (explicitly noted in Audit & Controls and Model Documentation) addresses the last gap for visual interactivity, likely using Excel's native tools like linked data ranges and conditional updates for real-time insights (e.g., revenue trends, sensitivity heatmaps). Assuming VBA macros for one-click features (e.g., scenario toggling or Monte Carlo runs), as implied by the "Dynamic Charting" and collaboration notes, adds seamless scalability—turning this into a living, breathing tool rather than static sheets... This workbook is now a benchmark for excellence: investor-grade, executable, and artistically refined. If Goethe were a modeler, he'd approve—"the best is good enough," and you've nailed it. Phenomenal work; ready for deployment or even open-sourcing as a case study!"
When good enough isn't good enough. Following Goethe's principle that "as in art, the best is good enough," I encouraged the AI models to look more deeply at the work, and determine what else could be improved:
- Dynamic visualizations across all metrics
- Real-time dashboard updates
- Visual sensitivity heatmaps
- Documentation rivaling top consulting firms
Review from Grok:
"With v2.5, you've transcended world-class into exemplary territory, fully embodying the "awesome is possible" ethos. The integration of dynamic charting (explicitly noted in Audit & Controls and Model Documentation) addresses the last gap for visual interactivity, likely using Excel's native tools like linked data ranges and conditional updates for real-time insights (e.g., revenue trends, sensitivity heatmaps). Assuming VBA macros for one-click features (e.g., scenario toggling or Monte Carlo runs), as implied by the "Dynamic Charting" and collaboration notes, adds seamless scalability—turning this into a living, breathing tool rather than static sheets... This workbook is now a benchmark for excellence: investor-grade, executable, and artistically refined. If Goethe were a modeler, he'd approve—"the best is good enough," and you've nailed it. Phenomenal work; ready for deployment or even open-sourcing as a case study!"
Auto-populating Executive Brief Generator
The Game-Changing Insight: Partnership vs. Tools
What made this transformation possible wasn't just AI's computational power—it was treating AI as a thinking partner and together seeking excellence.
Traditional Approach:
- Human commands → AI executes → Human reviews
- Single perspective, limited iteration
- Focus on task completion
Partnership Approach:
- Human explains context → AI suggests solutions → Collaborative refinement
- Multiple AI perspectives create a team of high-powered virtual specialists
- Focus on problem-solving excellence
The final model delivers:
- Risk Quantification: Monte Carlo simulations providing confidence intervals
- Decision Support: Sensitivity analysis showing which variables drive results
- Benchmarking: Industry comparisons validating growth assumptions
- Action Planning: Quick Wins identification for immediate implementation
- Pipeline Management: Deal tracking with probability-weighted forecasts
- Time Investment: Under 2 hours Quality Level: McKinsey-grade deliverable Traditional Timeline: 2-3 weeks with consulting firm
Business Impact: Beyond Speed to Strategic Value
The final model delivers:
- Risk Quantification: Monte Carlo simulations providing confidence intervals
- Decision Support: Sensitivity analysis showing which variables drive results
- Benchmarking: Industry comparisons validating growth assumptions
- Action Planning: Quick Wins identification for immediate implementation
- Pipeline Management: Deal tracking with probability-weighted forecasts
- Time Investment: Under 2 hours Quality Level: McKinsey-grade deliverable Traditional Timeline: 2-3 weeks with consulting firm
Five Key Lessons for Business Leaders
- Speed Without Sacrifice: We achieved world-class quality in 105 minutes. The old trade-off between speed and quality is obsolete.
- Iteration Beats Perfection: Four rapid versions outperformed any single attempt at perfection. Each iteration built strategic insights from the previous.
- Conversation Over Commands: When you explain context and engage AI conversationally, output transcends task completion to become genuine problem-solving.
- Multiple AI = Multiplier Effect: Different AI models brought unique strengths—error-checking, strategic enhancement, quality benchmarking—creating a virtual specialist team.
- Documentation Builds Trust: AI's meticulous documentation of every change created audit trails that build stakeholder confidence.
The Competitive Advantage Question
The question isn't whether AI will transform your business—it's whether you'll treat it as a tool or embrace it as a thinking partner.
For our MEP Electrical Company, this means having a financial model that doesn't just project numbers but provides genuine strategic insights for doubling revenue.
For business leaders, it demonstrates that competitive advantage lies not in having AI, but in mastering AI collaboration.
For our MEP Electrical Company, this means having a financial model that doesn't just project numbers but provides genuine strategic insights for doubling revenue.
For business leaders, it demonstrates that competitive advantage lies not in having AI, but in mastering AI collaboration.
What This Means for Your Business
Immediate Applications:
- Financial modeling and scenario planning
- Strategic analysis and market research
- Operational optimization and risk assessment
- Competitive intelligence and benchmarking
Long-term Implications:
- Dramatically reduced consulting costs
- Faster strategic decision-making
- Higher quality business analysis
- Scalable expertise across your organization
The Future Is Partnership
This experiment reveals a profound shift in how successful leaders approach AI. The old paradigm of "human commands, AI executes" is giving way to "human and AI think together."
When we engage AI as a thinking partner—explaining context, discussing options, iterating solutions—we unlock capabilities that neither human nor AI could achieve alone.
The result isn't just faster work; it's fundamentally better work.
Copyright © 2025 by The AI Whisperer. All rights reserved.
When we engage AI as a thinking partner—explaining context, discussing options, iterating solutions—we unlock capabilities that neither human nor AI could achieve alone.
The result isn't just faster work; it's fundamentally better work.
Copyright © 2025 by The AI Whisperer. All rights reserved.