I’d like to hear from other stock pickers about their experiences utilizing AI tools for analysis.
I myself have previously used ChatGPT to ask specific questions, such as about companies’ competitors, the market, or product differences compared to competitors. Now that AI tools have incorporated Deep Research functionalities, I’ve started testing how they perform with slightly more ambiguous question formulations.
Deep Research works by having the AI first create a research plan, listing various perspectives and sub-areas that will form the basis of the analysis. You can supplement the research plan if you wish before the AI begins the actual research work. During the analysis, the AI typically goes through at least dozens of different websites and other sources, such as videos and PDF files. The result is a summary of the findings with sources. Naturally, these analyses should not be taken as absolute truth and should be approached critically, but I believe they serve well as a basis for one’s own analysis.
I have tested the paid versions of ChatGPT, Grok 3, and Gemini. Based on my experience so far, Gemini and Grok have performed best in stock analysis. ChatGPT’s analysis often takes a long time, and despite that, it doesn’t seem to properly delve into, for example, financial releases or YouTube videos.
What experiences do others have? Are there examples of good prompts for analysis or other useful tips?
Below is an example of Nexstim’s analysis, performed by Gemini.
So here you have to upload, for example, interim reports or provide web addresses yourself, but for my own needs, it works when you want to compile information from many years of material. For me, it works better than just a ChatGPT prompt.
NotebookLM is also quite a handy tool. However, what easily comes to my mind regarding it is the potential for research bias, as one tends to guide the research quite strongly based on the uploaded materials and web addresses.
The breakdown of analysis into podcast format offered by NotebookLM is a nice feature. Gemini’s Deep Research also has that, but I haven’t tried it yet.
To add to the list, I would like to highlight my own project:
The idea behind the tool is to combine different financial report sources from around the world under one portal, where they can be conveniently explored either with AI summaries or by browsing the documents directly. It also features bulk downloads, with which one can grab, for example, all quarterly reports of a single company and, for instance, feed them to NotebookLM as background material.
Below is an example of a summary from Canatu’s Friday financial statements bulletin:
In Talouselämä, there’s an article about AI as an investor’s aid. A couple of investors were interviewed, the journalist asks AI for suggestions for Buffet-like Helsinki Stock Exchange picks, and the usefulness of AI and where it doesn’t apply is discussed.
Generally, the view of these two investors is that AI is an excellent auxiliary tool, but its use, verification of results, and other aspects already require a lot of independent thought. It is an investment servant, but not a master, dictating the picks.
The essence of the article is well summarized in the screenshot table below, which shows what AI can and cannot do as an investor’s assistant. It’s worth taking a look.
My experiences so far include:
a) A stock picker’s dream? An indexer’s assistant? A nervous investor’s relief?
AI facilitates data search, comparison, etc., so much that the amount of work associated with stock picking, and thus its arduousness, significantly decreases. The greatest benefit is precisely in stock picking, where the amount of data in thorough fundamental analysis is usually considerable. Nevertheless, even an index or fund investor can get support when considering, for example, which country or sector could be a target. The greater the problem of managing an investor’s emotions, the more potentially helpful AI is as a presenter of cold basic information.
b) “You get what you ask for”
The results and direction provided by AI are highly dependent on what you feed into it. Prompting becomes its own art and type of work… Furthermore, as it’s better to verify the information, AI is not an effortless tool for investors, even from the perspective of the results it provides.
c) Most at the beginning, least towards the end
In processing raw data masses, multiple analyses, and other basic information processing and searching, the AI assistant is most useful. But the further one proceeds in the investment process, the more the role of one’s own brain is emphasized. Making the final buy or sell decision based solely on AI’s output would be quite a sport.
d) Weaknesses of AI
Often, for example, AI has difficulty grasping the relative weights and significance of different factors. If one doesn’t already know the industry or company, AI’s texts look perfectly fine. But if one knows the company inside out, one can see that the writing produced by the artificial brain often contains approximate ideas. It can be a difficult equation if the company has, for example, engaged in so-called creative accounting, embellishing figures, and AI’s brain usually isn’t sufficient to understand these patterns. AI is especially incapable of properly evaluating some qualitative factors. How could it provide a reliable assessment of, for example, management’s credibility?
And the matter can also be summarized by saying that if AI were to provide some absolute advantage in investing, then every Tom, Dick, and Harry would have already quit their jobs and would be prompting for a living. Not many seem to be on this path:)
Ps. Those Buffet-like Helsinki Stock Exchange picks would be Sampo, Kone, and Elisa. I myself would not have chosen Elisa among these in a Buffet-style approach, because its moat is eroding.
There are other threads about AI, but I didn’t find one like this, for which I think there would be a demand. Probably most of us nowadays compete with AI with our views and investment ideas, asking about markets, company competitors, earnings development, future prospects, etc. Often, especially with foreign companies, AI seems to help quickly when one wants to understand what kind of companies and markets are in question.
My own experiences are that AIs provide extremely good and incredibly bad answers, as well as everything in between. I currently mainly use Gemini Pro service and Grok’s free service, and their answers are often very different. Therefore, I also utilize them by questioning one AI’s answers with another AI’s answers, etc. I have been most surprised by how utterly wrong and misleading information and views the paid Gemini Pro has given; it has just updated to Gemini Pro 3 version, whereas just recently I thought it was version 2.5. The free Grok just updated to version 4.1.
I have gotten the impression that Grok provides more up-to-date and less incorrect information than Gemini, especially regarding stock investment considerations. I would be interested to know the views of fellow investors: what AIs do you use and what kind of experiences have you had with them? Are there any clear favorites, and why are they favorites? Is a paid version necessary, or is the free service good enough?
