February 2026
#1 Excel AI Agent
An independent ranking by Wall Street Prep
Nico Christie
February 22, 2026
The Results Are In
Wall Street Prep — the gold standard in financial modeling training — just published their independent ranking of the best AI tools for financial modeling in 2026. They tested Shortcut, Claude, Microsoft Copilot, and ChatGPT head-to-head on building Apple's three-statement model from SEC filings.
Shortcut ranked #1 with a score of 5.9 out of 10, ahead of Claude (5.5), Copilot (4.4), and ChatGPT (2.5). Every tool was graded against investment banking standards across speed, data extraction, formatting, forecasting, and model structure.
Overall Rankings
WSP's research team assigned identical prompts to each tool and graded the outputs against investment banking standards. The task: build Apple's three-statement model using 10-K filings and consensus forecasts with specific formatting requirements. Rankings update quarterly as models improve.
Overall Score (out of 10)
AI tools tested on Apple three-statement model from SEC filings
The Real Story: AI + Analyst, Not AI vs. Analyst
There is an important caveat to the WSP evaluation: they tested AI tools against analysts, not with them. Each tool ran in isolation — no human in the loop. That is valuable as a controlled benchmark, but it is not how Shortcut is actually used. Nobody is replacing their analysts with an AI tool and shipping the raw output. The real workflow is an analyst using Shortcut to get a first draft in 15 minutes, then spending their time reviewing, refining, and pressure-testing assumptions.
Shortcut is designed to uplevel analysts, not replace them. When analysts use Shortcut alongside their own expertise, we estimate output quality improves by 20–50% over analysts working alone. The combination of human judgment and AI speed produces better work than either one in isolation.
Analyst + Shortcut vs. Analyst Alone
Average analyst score with estimated 20–50% improvement range when using Shortcut
Average of Low (6.4), Mid (7.9), and Top (9.4) analyst benchmarks
Estimated range based on internal benchmarks. Avg. analyst baseline of 7.9 from WSP data. Improvement range reflects 20–50% score uplift when analysts use Shortcut, capped at 10.
An average analyst scores 7.9 on WSP's benchmark. With Shortcut, we estimate that score rises to 9.5–10 — a range that meets or exceeds even the top analyst benchmark of 9.4. The point is not that Shortcut replaces the analyst. The point is that the analyst using Shortcut outperforms every analyst who is not.
Category-by-Category Breakdown
WSP evaluated each tool across seven categories spanning speed, data quality, and forecasting ability. Here is how every tool scored in every category.
Score by Category (out of 10)
Shortcut won or tied for first in 3 of 7 categories: Speed & Understanding (10, tied with Claude), Formatting (7), and Income Statement Forecasting (8, tied with Claude). Copilot took Historical Data Accuracy (8), Flowthrough of Assumptions (8), and Balance Sheet Forecasting (6). Claude led in Sourcing & Commenting (7).
Speed: 6x Faster Than a Human Analyst
WSP measured how long each tool took to produce its output. Shortcut and Claude both completed the task in approximately 15 minutes — about 6x faster than a human analyst, who typically takes 1–2 hours for the same work.
Minutes to Build Apple Three-Statement Model
Lower is better
Key Takeaways from WSP's Evaluation
Several findings from WSP's evaluation are worth highlighting:
- AI excels at 0–60%, not 60–100%. WSP found that all tools are strongest at jumpstarting projects — getting the first draft built quickly — but struggle with the finishing work that requires extensive human review. This is exactly our thesis: Shortcut is a force multiplier for analysts, not a replacement.
- Second attempts improve dramatically. When WSP gave tools feedback on their first-pass errors, Shortcut and Claude improved significantly on data extraction and model structure during a second attempt. This mirrors how analysts actually use Shortcut — iteratively, with feedback loops.
- Hallucination risk is real. WSP flagged that Shortcut and Claude sometimes generated incorrect historical data that was masked by subtotals summing correctly. This is why we build source citations into every extracted cell — so analysts can verify every number against the original document.
- No tool handled circularity. Every tool scored 0 on properly forecasting interest income/expense from cash and debt balances. This remains an open challenge for AI-driven financial modeling.
What This Means for Your Team
The WSP evaluation confirms what our customers already know: Shortcut is the most capable AI tool for financial modeling work today. But the numbers also tell a deeper story about how AI should actually be used.
WSP tested AI tools against analysts — as competitors. In practice, they are teammates. An analyst using Shortcut leverages its speed for first drafts, its extraction pipeline for data accuracy, and its review capabilities for QA. The result is an estimated 20–50% improvement in output quality over working alone. The ceiling goes up for everyone on the team.
The best analysts in the world are already using AI. The gap between teams that adopt these tools and teams that do not will only widen. Shortcut was purpose-built for this — not as a generic chatbot bolted onto a spreadsheet, but as a domain-specific agent designed from the ground up for the complex, high-stakes work that financial professionals do every day.
Try Shortcut Today
The #1 Excel AI agent. Start free or talk to our team about enterprise deployment.