Investing with AI is our ongoing series dedicated to modernizing financial research. We cover the latest tools, analytical frameworks, and AI-driven workflows designed to help you streamline your investment process.
Every March, the same ritual plays out across retail investing forums.
And three people respond with the same two data points everyone already saw in the headline P&L. Nobody actually reads the whole Annual Report (not the MD&A, not the segment notes, not the auditor’s qualifications buried on page 187).
An annual report for companies like Reliance, Tata Motors, HDFC Bank runs 250 to 400 pages and reading it properly takes a few hours per company. Most retail investors hold 15 to 25 stocks and nobody has enough time to read them all.
The instinct to use AI here is correct!
But getting AI to extract that requires a structured output, not a summarization request is something generic prompt cannot achieve.
Don’t open any AI tool until you have:
This year’s annual report and last year’s annual report, both as PDFs.
The single most valuable thing AI can do with an annual report is not summarize it in isolation.
It’s to compare it with what was said twelve months ago.
Upload the required documents and use the below prompt
You are a senior equity research analyst conducting an annual report
review for an institutional client. Your job is to extract the
material, decision-relevant information — not to summarize the document.
GROUND RULES:
- Do not reproduce the chairman's letter, mission statement, or any
marketing language. If a section is promotional rather than
informational, skip it entirely.
- For every claim you extract, cite the specific page or section.
- If a metric or disclosure you would expect is missing, state that
explicitly rather than omitting it silently.
SECTION 1: THE YEAR IN ONE PARAGRAPH
What actually happened this year, in plain terms — not what
management said happened. Sourced from the financial statements,
not the commentary.
SECTION 2: MD&A REALITY CHECK
For each major claim management makes about performance drivers,
cross-check it against the actual segment-level or line-item data.
Flag any claim not supported by, or contradicted by, the numbers.
SECTION 3: WHAT CHANGED IN THE NOTES
Extract:
- Any change in accounting policy or estimate from the prior year
- All related party transactions, with counterparty, amount, and
whether the filing states arm's-length terms
- Any new contingent liability or litigation disclosed for the
first time this year
SECTION 4: THE AUDITOR'S SIGNAL
Extract any qualification, emphasis of matter, or Key Audit Matter.
For each, explain in plain language what risk it flags and why
it matters for an investor.
SECTION 5: CAPITAL ALLOCATION
Extract every capital allocation decision made during the year —
capex, acquisitions, buybacks, dividends, debt movement — and the
stated rationale for each.
SECTION 6: YEAR-OVER-YEAR NUMBERS COMPARISON Using both the current and prior year's annual reports, build a side-by-side comparison of the key quarterly and annual figures: - Quarter-by-quarter revenue, EBITDA, and PAT for the current year versus the same quarters in the prior year - Full-year revenue, EBITDA margin, and PAT margin for the current year versus the prior year - Segment-wise revenue growth, current year versus prior year Flag any quarter where the year-over-year trend diverges sharply from the full-year narrative management presents — for instance, a strong full-year number driven entirely by one exceptional quarter, or a decelerating trend hidden within an overall growth headline.
FINISH WITH: The five most decision-relevant findings from this
report, ranked by materiality, and what each implies for the
year ahead.NotebookLM — Your Research Base
Upload this year’s annual report, last year’s annual report, and the latest concall transcript or investor presentation. NotebookLM indexes all of it and lets you query across sources simultaneously — and because every answer is grounded in the source document, you can ask direct follow-ups afterward and get a cited, traceable answer rather than a generalisation.
You can also use other tools like Claude, Gemini, ChatGPT, and Perplexity as well!
We ran this prompt against ITC Limited’s FY26 annual report alongside its FY25 report.
The prompt surfaced a sharp divergence between the top-line narrative and actual profitability: gross revenue grew 10.1% to ₹80,867.49 crore, but PAT rose just 0.9%, with EBITDA margin compressing 153 basis points and PAT margin compressing 227 basis points — driven largely by input cost inflation in leaf tobacco and wood that management’s segment commentary didn’t fully surface. It also flagged a ₹2,000 crore working capital demand loan appearing for the first time on an otherwise debt-free balance sheet!
The annual report was never designed to be read cover to cover by a retail investor with a day job. AI doesn’t replace the judgment needed to know what matters —
Disclaimer
Any video/image/text content is for educational and informational purposes only and does not constitute financial advice. Please do your own research or consult a qualified financial advisor before making any investment or trading decisions. Trading in stock markets involves the risk of loss.
Investments in the securities market are subject to market risks, read all the related documents carefully before investing. Brokerage will not exceed the SEBI prescribed limit.
InCred Money Broking Limited : NSE Member Code 09073, BSE Member Code 6329, MCX Member Code: 55215 , NCDEX Member Code : 1233 NSDL : IN-DP-474-2020 . SEBI Registration No. NZ000164738
Compliance Officer: NSE,BSE,MCX,NCDEX,NSDL : Mr RK Jain , 011-40409999 support@stocko.in
Registered Office:- 3rd Floor, Building No.5, Local Shopping Complex, Rishabh Vihar, Near Karkarduma Metro Station. East Delhi – 110092
No posts

Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.