Created an app to challenge and synthesise my investment thesis into assumptions and trackable metrics
My career background is in finance/fintech and I've often invested into concentrated value stocks that have served me well, but I felt like I didnt have enough time to do the process as I would like.
With the advent of LLMs I decided to create something that could perform the role of an Investment Committee in challenging the initial thesis and then an analyst in tracking agreed metrics against thresholds on a regular cadence.
So I created this investment thesis [app](https://investmentthesis.app/), totally free to use, you start with $2 of credits with the LLM and then you can put your own anthropic key to continue using. I make nothing from this and just wanted to share in case other people find it useful.
**What it actually does**
It runs your idea through the same process a professional firm would:
\- The interview. It interrogates your initial idea and defines it in terms of assumptions and metrics that test the assumptions hold. Every metric you agree to track gets its source verified against a real filing or report before it's accepted.
\- The bear case. It builds the strongest case against you and makes you defend it, one argument at a time. You can concede points, adjust the thesis, or hold your ground. You choose what to track and ultimately own the thesis.
\- Firm-level watches. The qualitative checks a good analyst runs that retail investors don't have the resource to do. Classic Buffett-style example: did management actually deliver what they promised in previous quarterly reports, or did the promised metric quietly disappear?
\- One commit. Nothing goes on the record until the end. When you commit, you commit to everything at once: the thesis, the assumptions, measurable tripwires ("if churn goes above X for two consecutive quarters, that's my exit signal") and the exit rule.
\- The monitoring page. This is the part that genuinely saves me time, found it hard in a project to get Claude to reliably do so without missing areas which is what led me to start working on this. It checks your tripwires against new filings on the agreed cadence while you're not looking, shows you a brief before earnings of exactly what your own falsifiers say to watch for, and when a tripwire fires it makes you deal with it honestly: accept it, challenge it with evidence, or admit you're overriding your own rule and record why. Holds you to account when otherwise you may allow yourself to drift, I certainly have in the past.
\- And if you can't articulate a real edge in the interview, it tells you to buy an index fund and saves nothing. That's deliberate and I hope keeps me honest!
It never tells you what to buy. The judgement is always yours, it just refuses to let you be vague about it. (Not financial advice.)
**The pain points that made me build it:**
\- I'd research for two weeks, buy, and then just watch the price. When it dropped I'd re-read the bull points I already agreed with.
\- My assumptions were never written down anywhere, so they could quietly shift and I'd never notice.
\- I never decided what would make me sell until I was already staring at a loss, which is the worst possible time to decide.
\- Monitoring properly takes hours per position per quarter. I have a job.
\- And when I tried doing this with a general chatbot, it was too agreeable. It wanted me to be right. I needed something designed to want me to be wrong.
On the trust question, because I'd ask it too: the key handling code is open source [here ](https://github.com/jeev13/thesis-key-audit)so you can read exactly what happens to your key (encrypted at rest, never shown again, there's also a "this session only" mode that never stores it). I'd advise creating a key in a dedicated Anthropic workspace with a spend limit regardless. A full thesis costs $3–5 in API costs.
I built this with zero real coding background using Claude Code, in about a month of spare time, which still feels absurd to me.
Happy to answer anything :)
TLDR — Free app that stress-tests your investment idea like an investment committee, makes you commit to exit tripwires in advance, then monitors them for you. $2 free go, then your own Anthropic key. Never tells you what to buy. [https://investmentthesis.app](https://investmentthesis.app)