AI and Financial Analysts: What's Actually Happening and What to Do

The honest assessment

Financial analysis is one of those professions where AI has gone from "interesting experiment" to "doing half my job" in about two years. That's not an exaggeration. i wish it were.

Here's what AI can do right now. Build financial models from raw data. Perform variance analysis and write coherent commentary about what's driving the numbers. Forecast revenue based on historical trends and market data. Analyse company filings and extract key financial metrics. Generate investment research summaries from earnings calls. Microsoft Copilot can sit inside your Excel spreadsheet and write formulas, spot anomalies, create charts, and explain what the data is showing... all in plain English. ChatGPT can take a set of financial statements and produce analysis that reads like a first-year analyst wrote it on a good day.

What it can't do, and this is where it matters, is apply judgement to uncertain situations. It can tell you that revenue is down 12% quarter-on-quarter. It can't tell you that the revenue decline is actually fine because the company is strategically exiting a low-margin product line that was dragging down profitability, and management has a credible plan to replace it within two quarters. That requires understanding context, reading management quality, and making a call about the future. AI is good at analysing the past. Investing is about the future.

The area shifting most rapidly is the sheer speed of analysis. What used to take a team of analysts a week... building a comp set, pulling financial data, constructing a model, writing the commentary... can now be done by one person with AI tools in a day or two. That compression is real and it's happening everywhere from boutique advisory firms to Goldman Sachs. JPMorgan's COiN platform processes 12,000 commercial credit agreements per year. That used to be 360,000 hours of lawyer and analyst time.

Your exposure level: High

High exposure. And i know that's not what you want to hear if you've spent years learning to build discounted cash flow models in your sleep.

The reason is straightforward. Financial analysis involves taking large amounts of structured and semi-structured data, applying quantitative methods to it, and producing written output that summarises findings and recommendations. Every single part of that pipeline is something AI handles well. Not perfectly. But well enough that the economics of financial analysis teams are changing.

Bloomberg's AI tools can now monitor markets, flag anomalies, and generate research notes. Morningstar uses AI for fund analysis. The big banks are all reducing analyst headcount in areas where AI can do the first pass. Graduate recruitment in certain financial analysis roles is already declining. If you're a junior analyst whose primary contribution is building models and writing reports... the window is narrowing.

The saving grace, if there is one, is that financial markets are adversarial. Everyone is looking for an edge. If AI gives everyone the same analysis (which it largely does, since it's trained on the same data), the edge comes from the human who can see what the model misses. The analyst who knows that a CEO's body language in the earnings call was off. The one who spotted a footnote on page 47 of the annual report that contradicts the headline numbers. That's still human work. For now.

The 90-day action plan

1. This week: let AI build your next model. Take a company you know well. Give ChatGPT or Claude the last three years of financial statements and ask it to build a three-statement model with assumptions. Compare it to what you'd build. Notice where the assumptions are lazy and where the structure is surprisingly competent. That comparison is educational.

2. Week two: automate your variance analysis. Open Copilot in Excel with your latest management accounts or financial data. Ask it to identify the top five variances and write commentary for each. Edit the output. If you can produce your monthly variance commentary in 30 minutes instead of three hours, you've made yourself more efficient and more valuable.

3. By day 30: use AI for competitive analysis. Pick a sector you cover. Use Perplexity to pull recent earnings data, analyst consensus, and market commentary for the top five companies. Use Claude to synthesise it into a sector overview. Compare it to the last sector report you wrote manually. Note the quality gap and figure out how to close it.

4. By day 45: build a prompt library for financial writing. Create saved prompts for your most common outputs. Quarterly commentary. Investment memos. Board-ready financial summaries. Include your house style, preferred metrics, and level of detail. A well-crafted prompt library means consistent, fast output that still sounds like you.

5. By day 60: go deeper on the qualitative side. The parts of financial analysis AI can't replicate well are the parts that require real-world understanding. Industry dynamics. Management quality assessment. Competitive moat analysis. Pick one and become genuinely expert. Read the annual reports AI can't be bothered to read. Talk to the people AI can't call.

6. By day 75: present AI-augmented analysis. Take your next piece of research and produce two versions. One traditional. One AI-assisted. Show your manager or team the time difference and quality comparison. Be honest about both. This demonstrates maturity and practical understanding.

7. By day 90: decide your direction. Financial analysis is forking. One path leads to highly technical quantitative work where AI is a power tool and the analyst adds deep domain expertise. The other leads to strategic and advisory roles where the analysis is a means to an end. Both are viable. Sitting in the middle isn't.

AI tools you should be using this week

• Microsoft Copilot for Work — This is the big one for financial analysts. It sits inside Excel and does things that would make a 2020 analyst weep. Formula generation, data analysis, anomaly detection, chart creation, and plain-English explanations of complex datasets. If you're still writing VLOOKUP formulas manually, you're doing it the hard way.

• ChatGPT for Work — Excellent for drafting investment memos, earnings call summaries, and financial commentary. Feed it data and it writes. Not perfectly, but fast. Also useful for brainstorming investment theses and stress-testing assumptions by asking it to argue the other side.

• Claude for Work — Better for longer analytical work. Paste in an entire annual report and ask specific questions. It handles large documents without losing track of details, which matters when you're doing due diligence or comparing multiple company filings.

• Perplexity for Research — Pulls real-time information with citations. Use it for market data, competitor analysis, recent earnings commentary, and macro trends. Faster than manual research and the citation feature means you can verify everything.

What to say in meetings

In your next team meeting: "I've been using Copilot to automate our variance analysis process. It cuts the time by about 60%. I'd like to use that time to go deeper on the qualitative factors that actually drive our recommendations." That's not threatening to anyone. It's practical.

When a senior analyst or portfolio manager asks about AI: "The tools are good enough for first-pass analysis and data work. The edge is in the interpretation. I'm using AI to get to the interpretation faster." That shows you understand both the opportunity and its limits.

If someone suggests AI could replace the analysis team: "It can produce the analysis. It can't tell you why the CEO's growth targets are unrealistic based on supply chain constraints that aren't in the filings. That's still us." Be specific. General claims are easy to dismiss. Specific examples stick.

If the worst happens

If you're made redundant from a financial analysis role, your skills are highly transferable. Financial modelling, data analysis, written communication, and the ability to form views under uncertainty are valuable in corporate finance, management consulting, private equity, venture capital, and corporate development roles. The core skill of "looking at messy information and telling people what it means" is one of the most broadly applicable professional capabilities there is.

Natural adjacent moves: corporate development, FP&A, management consulting, investor relations, or fintech product roles. The fintech space in particular is hungry for people who understand both finance and technology. If you can demonstrate AI proficiency alongside traditional financial analysis skills, you're a candidate that companies struggle to find.

The freelance and contract market for financial analysts is also viable. Companies need modelling work, due diligence support, and financial research for specific projects. An analyst who can do the work of three because they've integrated AI tools effectively... that's a consultant who can charge accordingly. The key is framing it right. You're not a displaced analyst looking for work. You're someone who understands modern financial analysis, including the tools, the methods, and the limitations. Different story. Different outcome.