How to Research a Stock With AI: A Step-by-Step Guide

Federico Kunze Küllmer is the founder of Moneda, one of the tools compared here.

Short answer: Ask the AI to explain the business, summarise the latest filing and what changed, check the numbers and the quality of earnings, put the valuation in context, judge the competitive position, lay out the bull and bear case, and test how the stock fits your portfolio. Check that every figure has a source. Then make the call yourself.

Researching a stock used to mean an evening with a stack of reports. AI gets you a first draft in minutes: it reads the filings, pulls the numbers together and explains them in plain words. You still have to ask good questions and check what comes back.

Below are eight steps. For each one you get a prompt to copy, a description of a good answer, and the thing you should check yourself. They work for any stock and most AI tools.

Step 1: Understand the business

Start simple. If you can't explain how a company makes money in two sentences, you're not ready to judge its price.

What to ask:

Explain what [company] does in plain English. How does it make money?
Which products or regions bring in most of the revenue?
Who are its customers?

A good answer contains: the main revenue streams with their share of total sales, the type of customer (consumers, businesses, governments) and how the company gets paid (one-off sales, subscriptions, fees).

Verify yourself: the revenue split. Companies report it by segment in their annual filing. If the AI gives percentages, ask where they come from.

Step 2: Read the latest filing and what changed

US public companies file an annual report called a 10-K with the Securities and Exchange Commission (SEC). The SEC describes it as a detailed picture of the business, the risks it faces and its financial results for the year. There's also a shorter quarterly report, the 10-Q. These filings are the primary source. News articles and analyst notes are commentary on them.

The most useful question isn't "what does it say" but "what changed."

What to ask:

Summarise [company]'s latest 10-K. What changed compared with last year's?
Which new risks were added? Did management change its guidance?

Guidance is what management says it expects for future sales or profit. A change in guidance often moves a stock more than the results themselves.

A good answer contains: specific changes (a new risk factor, a segment that grew or shrank, a revised outlook), each tied to the filing and its date.

Verify yourself: open the filing and read the "Business" and "Risk Factors" sections, plus management's discussion of results. The SEC's own guide suggests the Business section as a good place to start.

Step 3: Check the numbers and the quality of earnings

Now look at the financials. Three numbers tell you a lot:

• Revenue growth: is the company selling more each year?
• Gross margin: the share of each sale left after the direct cost of making the product. A 60% gross margin means 60 cents of every dollar remain to pay for everything else.
• Free cash flow: the cash left after running the business and paying for equipment and buildings. It's harder to dress up than profit.

Quality of earnings asks whether reported profit is backed by real cash. If profit keeps rising but cash flow doesn't, find out why.

What to ask:

Show [company]'s revenue, gross margin, net income and free cash flow
for the last five years, with the source for each figure.
Does free cash flow track net income? Flag anything unusual.

A good answer contains: a table with years, figures and sources, plus a short note on trends and any gap between profit and cash.

Verify yourself: spot-check two or three figures against the filing. If one is wrong, treat the rest with care.

Moneda AI, the agent inside Moneda, runs this quality of earnings check as part of its research and shows the source for each figure.

Step 4: Put the valuation in context

A great company can still be a poor investment at the wrong price. Valuation tells you how much you pay for what you get.

The most common measure is the P/E ratio (price to earnings): the share price divided by the profit per share. A P/E of 20 means you pay 20 dollars for every dollar of yearly profit. On its own, the number means little. It needs context.

What to ask:

What is [company]'s current P/E and price to free cash flow?
How do they compare with its own five-year average and with its closest peers?
What growth does the current price seem to assume?

A good answer contains: current multiples from live market data, a comparison with history and peers, and a plain statement of what the price implies.

Verify yourself: the date of the price. Valuation goes stale fast. A P/E from memory or from last quarter can be badly off.

Step 5: Judge the competitive position

A moat is anything that protects a company's profits from competitors, like the water around a castle. A wide moat lets a business keep earning high returns for years.

A useful framework is the 7 Powers, from strategist Hamilton Helmer. Each power is a lasting edge:

1. Scale economies: costs per unit fall as the company gets bigger, so rivals can't match its prices.
2. Network economies: the product gets more valuable as more people use it.
3. Counter-positioning: a new business model that incumbents won't copy because it would hurt their existing business.
4. Switching costs: leaving is costly or painful for customers, so they stay.
5. Branding: people pay more because they trust the name.
6. Cornered resource: exclusive access to something valuable, such as a patent, a licence or rare talent.
7. Process power: a way of working so refined that rivals can't copy it quickly.

What to ask:

Score [company] on each of the 7 Powers. For each one, give the evidence
and say whether the power is getting stronger or weaker.

A good answer contains: a score per power with evidence, not just adjectives. "Strong network effects" means little without numbers like user growth or retention.

Verify yourself: whether the evidence holds up. Most companies have one or two real powers at most. If the AI awards all seven, push back.

Moneda AI scores companies on the 7 Powers and explains each score in plain language.

Step 6: Hear both sides of the story

Every stock has a bull case (why it could go up) and a bear case (why it could go down). If you only ask for the good news, you will usually get it, and you will learn very little.

What to ask:

Give me the strongest bull case and the strongest bear case for [company].
For each, what would have to happen, and what evidence would prove it wrong?

A good answer contains: two cases argued with equal effort, each with specific drivers and clear signs to watch.

Verify yourself: that the bear case is real, not a strawman. If you can't name what would make you change your mind, you haven't finished this step.

Moneda AI writes both cases side by side, and you can set alerts on the signals that matter to you.

Step 7: See how it fits your portfolio

A stock doesn't live alone. It lives in your portfolio, next to everything else you own.

Three checks help:

• Concentration: how much of your money would sit in this one stock, or in one sector or country?
• Benchmark: how has the stock done compared with something simple, like an index fund or gold?
• Stress test: how would your portfolio have held up in past market falls?

What to ask:

If I add [company] at 5% of my portfolio, how does my sector and country
exposure change? How would my portfolio have done in 2008, 2020 and 2022?
How has [company] performed against an S&P 500 index fund over five years?

A good answer contains: your exposure before and after, a benchmark comparison over the same period, and stress test results in plain numbers.

Verify yourself: that the AI uses your real holdings, not a guess.

In Moneda, the agent can stress test your portfolio against 2008, 2020 and 2022, and benchmark it against gold, an index or a fund.

Step 8: Decide yourself

By now you know the business, the numbers, the price, the moat, both stories and the fit. The decision is yours. AI can help you think it through, but it shouldn't make the call.

Write down three things before you act: why you're buying, what would make you sell and how much you're willing to put in. When the price moves, you will have something to check your reasoning against.

Moneda AI never tells you to buy or sell. If you decide to invest, you can do it in the same app, and every order needs your approval with your passkey.

How to tell if an AI's answer is reliable

AI answers sound confident whether they're right or not. Three checks help you tell research from guesswork.

Sources are cited. Every figure should link back to a filing or a market data feed, with a date. No source, no trust.

Figures come from data, not memory. General chatbots answer from what they learned in training, so they can state old or wrong figures with total confidence. Dedicated research tools fetch figures from filings and live market data at the moment you ask.

It says when it doesn't know. A reliable tool tells you when data is missing or unclear instead of filling the gap.

Moneda AI is built on Numa 1, Moneda's financial AI engine. It researches stocks, ETFs, a few commodities through ETFs, and crypto, from filings and live market data, with the source behind each figure. You can search a company's SEC filings and earnings-call transcripts directly, and a stock screener turns a plain-words idea into a list to research. In Moneda's own tests in October 2026, Numa 1 scored 97.9% on groundedness (answers supported by the source data) and 92% on claim verifiability (claims that can be checked against a cited source). Those are Moneda's own results, and you should still spot-check what matters to you. What sets Moneda apart is that research and investing live in the same app: answers show the source behind every figure, and if you decide to buy, every order needs your approval. One limit to know: you can invest only in US-listed shares and ETFs, so no European-listed stocks or funds.

FAQ

Can AI pick stocks for me?

AI can research a stock thoroughly and fast, but it shouldn't make the decision. Use it to understand the company and test your thinking. The choice stays with you.

Is ChatGPT good enough for stock research?

General chatbots are good at explaining ideas. For figures, check whether the tool pulls live data and cites sources. If it answers from memory, verify every number.

What is the most important number to check?

There isn't one. Free cash flow is a strong start because it shows the real cash a business produces. Read it alongside growth, margins and valuation.

I don't have a stock in mind. Can AI help me find one?

Yes. Describe what you're looking for in plain words, such as "profitable software companies growing over 15% a year," and a good tool will suggest ideas to research. In Moneda, the agent turns that kind of description into a list of ideas.

What is the Moneda Ratio?

A descriptive scorecard from 1 to 7 that sums up a company's profile in one view. It describes. It doesn't tell you what to do.

How long should researching a stock take?

With AI, a first pass through all eight steps can take under an hour. Take longer on the steps where your answers are least clear.

---

This guide is for education only and is not investment advice. Investing involves risk, and you can lose money.