($) Manus vs Gemini vs GPT: the Junior Stock Analyst
This post is for our premium members only. If you are reading this in full, thank you for being an Interconnected Premium member! If you aren’t, I hope you become one by scrolling down and tear down that paywall! 😎
Over the weekend, Manus, an AI agent product made by a two-year-old startup from China took over AI twitter with its product launch video and viral influencer marketing campaign. It is an invite-only product for now, and I managed to get an invite code to play with it.
Instead of writing a post on the company’s history, technology, and larger implications for the US-China technology competition (which I will do at some point), I’m doing a utilitarian “product review” post instead.
Why? Because I’m in the market for a good AI agent to help me do research!
As many of you know, I recently launched my own, long-only fund investing in the picks and shovels of the AI economy. Because it’s mostly a one-person show, I find the timing of running a fund in this era particularly exciting. Not only does my portfolio benefit from AI’s long-term impact, but I can also leverage the technology to maximize productivity in the here and now, without adding headcount. The timing couldn’t be better; it is a fund manager’s dream! (If you are in a similar boat, where your work is a lot of research and information synthesis, you may find this post particularly useful.)
So can Manus or another AI agent serve the role of a junior analyst?
To answer that question, the AI agents that I put in the arena for this exercise are Manus in standard mode, Gemini in Deep Research mode, and GPT 4.5 with Deep Research.
The task I chose for each AI agent is to do a deep dive research report on the cybersecurity company, CyberArk ($CYBR), and give me an investment recommendation. For me, this company makes for a perfect test case, because it is a company I’m genuinely interested in (no fake task), have some knowledge about it, but still need to learn more. Thus, I can both benefit from this exercise directly and spot mistakes that someone who doesn’t (loosely) follow the company may not be able to.
I gave each this exact same prompt: “Can you please create a deep dive research report on the company CyberArk, including the company's history, management, culture, M&A, financial details, and future business outlook? In the end, provide your recommendation on whether this company is a good long term investment opportunity at this moment in time. thank you.”
I wanted to see if these AI "analysts" could:
Actually understand what CyberArk does (harder than it sounds)
Provide detailed and accurate historical and financial information without hallucinating numbers
Produce a well structured research report with a clear, transparency process
Make an investment recommendation that may or may not render a real junior (or even senior) analyst obsolete
Here is my grading. (I have no dog in this fight, just telling it like it is.)