Thesis
Stock Ideas, Researched For You
I built a tool that turns an investing idea into a real shortlist of stocks. Looking for honest feedback.
Hey IH, first post.
I'm a AI engineer building a side project alone. Quick rundown of what it does, then a few questions.
You type an investing intent in plain English. Something like "data centers" or "GLP-1 winners". Most tools either give you a generic screener or send you to ChatGPT, which makes up tickers and forgets the answer the moment you close the tab. I wanted this for myself, to help me found new gems in my investing journey.
My tool does something different. It splits your idea into a few sharper angles. "Data centers" becomes four separate theses: the REIT angle, the hyperscaler angle, the power and cooling angle, and the picks-and-shovels angle. You pick the one or ones you actually want to research. For each, you get a ranked shortlist of real companies, with a clickable source on every claim (10-K, earnings call, news article, S&P report). Every thesis lives at a permanent URL so you can come back to it, share it, or send it to a friend.
What's live right now: 5 fully researched demo theses you can poke at (data centers, onshoring, GLP-1 second-order effects, energy, aging economy), 26 sub-theses across them with ranked companies and real sources, and a landing page with a waitlist if you want updates.
Link: https://try-thesis.com/
Hit one of the demo cards and click into a company. That is the experience I am trying to nail.
What I want feedback on:
1. First impression of the home page. Does the value prop land in 10 seconds, or do I need to rewrite it?
2. Click into one of the 5 demo theses. Are the per-company "why this fits" notes actually useful, or do they read like generic AI sludge?
3. The split into sub-theses (REITs vs hyperscalers vs power) is the move I'm betting on as the differentiator. Is that obviously useful to you, or does it feel like extra clicks?
Brutal honesty very welcome. Better to fix it now than after months of building.
4 Comments
Generic AI responses usually fail investors because they lack the specific financial sources needed to verify a thesis before putting real capital at risk. A great insight is that professional researchers often look for the bear case just as much as the bull case to understand the actual risk-to-reward ratio of a specific sub-thesis. Have you considered adding a feature that highlights the potential downsides or risks for each ranked company to give a more balanced view?
Go check it out and see for yourself :)
It is a solid approach because professional investors often spend more time trying to "kill" a thesis than trying to prove it right, so seeing those clickable sources makes the validation process much faster.
The sub-thesis split is the real winner here because an "AI play" is too broad to be actionable, but separating the power grid constraints from the chip manufacturers actually gives you a specific strategy to trade.
I focus on this type of deep positioning in my high-tier PR and media placement work where we find the specific "angle" of a company's story that will actually resonate with major news outlets.
Are you finding that users are gravitating toward the "second-order" effects more than the obvious first-tier companies?
HI, i'm Connell, a Bug Bounty Hunter. I run PentraSec, a Web Application Security boutique, similar to platforms like HackerOne and Bugcroud but more personalised, meaning more focus on in-scope applications to test.
I'd be happy to partner with you to test your applications before attackers do; leaving you to focus on the development while i cover your unintentionally open paths and footprints attackers could crawl into, and making sure you stay trusted and guaranteed by your clients.