About BeliefState
BeliefState gives your AI the context behind an investment idea: who said it, why they believed it, and what happened next.
We're building for anyone researching public trade ideas, especially retail and quantitative traders. MCP and the API bring the reasoning, evidence, and available outcomes directly to your AI.
Why I started BeliefState
I started BeliefState because I was frustrated by investment ideas being sold as advice. People were asked to pay for a prediction and hope it was right, without a clear way to judge the reasoning or follow what happened next. I wanted to make that record easier to check.
I want agentic trading to become accurate and dependable enough that people can feel confident relying on their own trading agent. That confidence has to come from evidence and measured outcomes. My goal with BeliefState is to make investing easier by bringing useful research directly to people's AI through MCP and the API.
Built by Ant
I'm a builder and creator at heart. I love AI and finance, and agentic finance brings those interests together. I build BeliefState independently and use it for my own stock research.
What your AI gets
Our goal is to help your AI assess likely outcomes for trade ideas and give you a stronger basis for conviction. That starts with the reasoning behind a view, its sources, and the outcomes we can measure.
The reasoning behind an idea
See who holds a view and the research behind it. When the source includes them, the record captures expected catalysts, risks, timeframes, and what would change the investor's mind.
What changed, and when
Follow recorded changes without losing the earlier view. The record separates when an idea was published, when BeliefState first found it, and when a change was recorded.
What happened next
Compare the idea with later price movements and benchmarks where market data is available. Losses and unresolved outcomes remain visible, along with gaps in the history.
What we stand for
Conviction should come from evidence you can check. Useful research should make the reasoning, risks, and results easier to understand.
Show the source
You should be able to check what supports an idea. Records link to dated evidence. If we use another publisher's summary, we label it and distinguish it from the investor's own words.
Be honest about outcomes
Published opinions, self-reported trades, and verified executions stay separate from calculated market returns. A winning call alone does not prove repeatable investing skill.
Say what's missing
We work with authorized sources and the stocks in our current coverage. Missing history and unresolved outcomes stay explicit, so you can see where the record is incomplete.
Use BeliefState
Bring the research to your AI
Connect a supported AI client through MCP, or use the API in your own software. Both give you access to the same research. Check which stocks and sources we cover before choosing a plan. Connection and coverage discovery are free. Research uses the connected account's available plan allowance or eligible prepaid credits. For access-denied requests, confirm the connected account and its current access. Return to the same AI chat and retry once access is available.