KEY TAKEAWAYS
- Work in order: identify the company, read financial history, compare peers, check ownership, read disclosures, then check data freshness.
- Name the tool or data you want in the prompt, and tell the AI to leave gaps blank instead of estimating.
- Use the same fields and periods for every company you compare, so differences are real and not definitional.
- DalalOS returns sourced facts and mechanical ratios only. Quote-style prices are end-of-day, and nothing here is investment advice.
Before you start
Connect DalalOS to Claude, ChatGPT or another MCP client first; the connect page has the current steps for each client, and this tutorial covers the general workflow. The examples use a large, well-covered company, but the same steps apply to any NSE/BSE-listed company the data covers.
The six steps and the tools behind them
| Step | Question | DalalOS tools |
|---|---|---|
| 1. Identify | Which company exactly, and what does it do? | search_stocks, get_company_profile |
| 2. Financial history | How have revenue, margins, returns and debt moved? | get_financials, get_financial_trends |
| 3. Peers | How do comparable companies look on the same fields? | get_peer_comparison, get_relevant_peers |
| 4. Ownership and pledge | Who owns it and is promoter pledge changing? | get_shareholding, get_pledge_trend |
| 5. Disclosures and valuation history | What did the company say, and how did valuation ratios evolve? | search_company_disclosures, get_valuation_history |
| 6. Freshness | How current is each block of data? | get_data_freshness |
Step 1: Identify the company
Start by resolving the name to a listed entity so every later call uses the same symbol. Many companies have similar names, and a wrong match quietly poisons the rest of the analysis.
EXAMPLE PROMPTS
- Use DalalOS search_stocks to find Asian Paints. Show the NSE symbol, BSE code and ISIN, then use get_company_profile to summarise its sector, size and latest reported fundamentals. Show the source and freshness date.
Step 2: Read the financial history
Look at several years of reported figures, then at direction. The levels tell you where a company is; the changes tell you where it is heading. Ask for consistent periods and for missing values to be flagged.
EXAMPLE PROMPTS
- Use get_financials for annual figures over the last five years: revenue, operating margin, net profit, ROE, ROCE and debt-to-equity. Use consistent periods, flag any missing field instead of estimating it, and cite the source.
- Use get_financial_trends to show year-over-year change in revenue and net margin, and say whether the latest quarter changes the direction. Show the underlying numbers.
Banks, NBFCs and insurers need their own fields
The financials tool adapts its output for lenders and insurers. For these, ask for the sector-relevant fields rather than industrial ratios such as debt-to-equity, and compare lenders only with lenders.
Step 3: Compare with peers on identical fields
A number means little on its own. Ask for peers and the same fields on the same dates, and ask the AI not to rank them: a ranking is a verdict you did not ask for.
EXAMPLE PROMPTS
- Use get_peer_comparison for Asian Paints and show PE, PB, ROE and ROCE for it and its peers, with the peer median. Do not rank or rate them; show the figures and the date of each.
- Use get_relevant_peers to pick a smaller operating peer set and explain why each peer was included, then compare the same fields.
Step 4: Check ownership and promoter pledge
Quarterly shareholding shows whether promoters, foreign institutions and domestic institutions are adding or reducing stakes. Pledge shows how much of the promoter’s stake is pledged against borrowings. Both are reported facts; the interpretation is yours.
EXAMPLE PROMPTS
- Use get_shareholding to show promoter, FII, DII and public holding for the last eight quarters, then get_pledge_trend for promoter pledge over the same period. Show the raw percentages and which quarters changed most.
Step 5: Read disclosures and valuation history
When a number or an ownership change looks interesting, go to the company’s own words. The disclosure search returns matched passages from indexed filings with original page references, so you can open the source. Valuation history gives point-in-time ratios that use only filings public on each date.
EXAMPLE PROMPTS
- Use search_company_disclosures for this company with the phrase 'capacity expansion' and show the matched passages with their source page references. Separate what the filing states from anything you infer.
- Use get_valuation_history to show monthly point-in-time PE for the last five years. Report the figures and the dates only; do not say whether the current level is high or low.
Disclosure search covers filings that have already been indexed, and filing tools can queue processing and use on-demand quota, so a filing that is not yet extracted may take a moment or be reported as a coverage gap. See data sources and freshness.
Step 6: Check freshness before you rely on anything
EXAMPLE PROMPTS
- Use get_data_freshness and tell me, for prices, financials and shareholding, when each was last refreshed and whether any is stale. Then list which of the figures above depend on a stale domain.
Put the pack together, then form your own view
Ask for a neutral summary
EXAMPLE PROMPTS
- Summarise the evidence above as three lists: what the numbers show, what is missing or stale, and what I should verify in the original filings. Do not give a rating, target price or buy/sell opinion.
Where AI analysis goes wrong
DalalOS is a research and data layer, not investment advice. Verify material figures against the original filings and make your own decisions, or consult a SEBI-registered adviser. To screen the market before you analyse individual companies, try the stock screener.
FAQ
Common questions
- Can AI do fundamental analysis of Indian stocks?
- AI can gather and organise the evidence quickly: financial history, peer comparisons, ownership and filings. With a connected data source such as DalalOS it works from sourced figures rather than memory. The judgement about what the evidence means is still yours, and DalalOS does not give ratings, target prices or buy/sell calls.
- Which DalalOS tools does fundamental analysis use?
- Typically search_stocks, get_company_profile, get_financials, get_financial_trends, get_peer_comparison or get_relevant_peers, get_shareholding, get_pledge_trend, get_valuation_history, search_company_disclosures and get_data_freshness. The full list is on the tools page.
- Does it work for banks and NBFCs?
- The financials tool adapts its statements and ratios for banks, NBFCs and insurers, so ask for the fields that matter for the sector, such as net interest income and asset quality for a lender, and compare banks with banks.
- How do I stop the AI from guessing?
- Tell it to use only the connected tools, to leave missing fields blank rather than estimate them, and to show the source and freshness date for every block of numbers. Then open the original filing for anything material.
- Is this investment advice?
- No. The guide shows how to collect and read facts and mechanically-computed ratios. It does not say whether any stock is cheap, expensive or worth buying, and prices are end-of-day rather than real-time. Verify the data and decide independently.
Connect your AI to Indian stock market data
Add DalalOS as a connector and sign in once. Free to start with no card and no waitlist; paid plans from ₹199 a month when you need more.
KEEP EXPLORING