Most people who try using AI for investing ask it things like “Is Apple a good buy right now?” and then wonder why the answer is useless. The problem is not Gemini. It is the question.
Gemini is not a stock picker. What it is genuinely good at is processing large amounts of information quickly, finding patterns across financial data, and helping you think through ideas you would not have reached on your own. The difference between a useful AI session and a waste of time comes down entirely to how you set it up and what you ask. Here are three methods I have been using, with the actual prompts.
Before You Start: One Setup Step Worth Doing
Before any stock research session, tell Gemini what kind of investor you are and what you care about. Something like this:
“I am a long-term investor focused on US equities. I care about revenue growth trends, profit margins, competitive positioning, and whether a business is run by management that allocates capital well. For any company we discuss, lead with the most important insight first, use specific numbers, and tell me when data is uncertain or when you are making an inference rather than citing a fact.”
This single step changes the quality of every answer you get afterward. Without it, Gemini defaults to generic textbook responses. With it, you get answers structured around what you actually want to know.
Method 1: Company Analysis That Goes Beyond the Surface
The traditional way people research a company is to look at a few financial ratios, read the latest earnings summary, and decide based on whether the recent quarter was good or bad. This approach misses almost everything that matters for long-term investing.
When I research a company using Gemini, I use a structured prompt that covers the things most investors skip:
“Analyze [Company Name] as a long-term investment. Cover these in order: (1) What does this company actually do to make money, and is that business getting stronger or weaker over the past three years? (2) Who are the real competitors, and what does it take to win in this industry? (3) What are the last three years of revenue growth, operating margin, and free cash flow trends? (4) What could go wrong that is not already obvious from the stock price? (5) What would have to be true for this to be a significantly better business in five years than it is today?”
Questions four and five are where this gets useful. Most summaries tell you what the business does well. The question about what could go wrong forces Gemini to surface non-obvious risks, things like regulatory exposure, technology shifts, or customer concentration. The question about what has to be true in five years tells you what the bull case actually depends on.
One important note: always verify the numbers Gemini gives you against actual SEC filings or financial data sites. Gemini can and does get financial figures wrong, especially for smaller companies or historical data. Use its analysis as a framework, not as a source of record.
Method 2: Expanding Your Investment Ideas in Directions You Would Not Think Of
This is the method most people do not know exists, and the one I find most consistently useful.
Instead of asking Gemini to analyze a stock you already have in mind, you describe a theme or trend you believe in and ask it to map out the full investment landscape around that theme:
“I believe that AI infrastructure spending will continue to grow significantly over the next several years. Map out the full investment landscape around this theme: (1) Who are the direct obvious beneficiaries, and what are their limitations? (2) Who are the less obvious second and third-order beneficiaries that most investors overlook? (3) Are there companies that would be hurt by this trend worth watching as names to avoid? (4) What are the ETFs covering this theme and what do they actually hold? (5) What would cause this entire thesis to be wrong?”
What you get back is not a stock pick. It is a map. You can see the obvious names everyone already owns, the less-covered companies worth researching further, and the structural risks to the whole theme. This is genuinely difficult to build on your own without spending hours reading research reports.
I ran this on a theme a few months back and ended up looking into a company I had never heard of before. It was up about 16% over the following three months. The AI did not pick that stock. It surfaced something I then went and researched properly myself. That is the distinction that matters.
Method 3: Building a Personal Investment Checklist
This method takes the most setup but pays off the longest. The idea is to use Gemini to build a consistent framework for evaluating every investment, so you are not making decisions differently each time based on how you happen to be feeling that day.
“I want to build a personal investment checklist that I apply consistently to every stock I consider buying. Based on long-term value investing principles, help me create a checklist with three sections: (1) Business quality — what I need to confirm about the durability and competitive position of the business, (2) Financial health — the specific numbers I should check and what thresholds matter, (3) Valuation — how I should think about whether I am paying a fair price. For each item, tell me what a good answer looks like versus a bad answer.”
Once you have the checklist, use it every time. Then apply it to specific companies:
“Apply this checklist to [Company Name] and give me a score of Pass, Fail, or Uncertain for each item, with a one-sentence explanation. Where you are uncertain, tell me specifically what I would need to research further to get a definitive answer.”
The “Uncertain” category is often the most valuable part. It tells you exactly where to focus your own research rather than trying to verify everything at once.
The Thing That Cannot Be Skipped
Gemini makes mistakes. It cites incorrect financial figures. It occasionally presents outdated information as current. It can sound very confident while being completely wrong on specifics.
Every number Gemini gives you needs to be verified before you make a decision based on it. This is not a reason to avoid using it. The frameworks, the questions it helps you ask, and the landscape-mapping capability are genuinely useful. But the final research step of checking actual data has to be yours.
The people who use AI well for investing treat it the way a good analyst treats a research assistant — useful for structuring the problem and surfacing ideas, but not the last word on anything specific. The judgment call stays with you. So does the responsibility for whatever decision you make.
Leave a Reply