How TRUE builds an answer.
If you're going to rely on a research tool, you're entitled to know exactly how it works — including where it falls short. This page is that disclosure.
For research and education. Not financial advice.
The pipeline
Question in, evidence out.
Every brief follows the same discipline, whether the question takes one second or one minute to answer.
- 1 Understand what's actually being asked, and how much depth it needs
- 2 Pull the relevant live data — prices, fundamentals, filings, macro, news
- 3 Reason over it, and check the story against what the data actually did
- ✓ Write the short read, state the uncertainty, and cite every claim
What data TRUE uses
- Market data — prices, volumes and historical series for stocks, ETFs, indices, FX and commodities.
- Fundamentals and filings — company financials and official disclosures.
- Economic data — the macro calendar and official statistical releases.
- News — established market news providers, treated as a claim to be checked rather than a fact to be repeated.
How it reasons
A question is routed to the depth it needs. A price lookup is answered directly from data. A "why did this move" question pulls context and cross-checks it. A deep research question reads filings and competitors before it says anything at all. Where a question needs more thought, TRUE takes more time — and shows you what it's reading while it does.
How it cites
Factual claims carry a source. If TRUE cannot support a statement with data, it is required to say so rather than fill the gap with plausible-sounding prose. "I don't know" is an acceptable output.
How it handles uncertainty
Every brief contains an explicit uncertainty section. This is not a disclaimer bolted on at the end — it is the part we consider most important. Where the evidence is genuinely mixed, TRUE says the evidence is mixed instead of manufacturing a conclusion.
Known limitations — stated plainly
We would rather tell you these than have you discover them.
- It cannot predict prices. Nothing can. TRUE explains what has happened and what is currently driving an asset. It does not forecast, and any output that reads like a forecast should be treated with suspicion.
- Language models can err. Grounding in live data and citation dramatically reduces fabrication, but does not eliminate the possibility of a mistake. This is precisely why every claim is sourced — so you can check rather than trust.
- Data can be delayed or wrong. Providers have outages and revise figures. Official statistics are revised routinely.
- It doesn't know you. TRUE has no knowledge of your circumstances, tax position, time horizon or risk tolerance, and therefore cannot assess whether anything is suitable for you. That is what a qualified financial adviser is for.
- Coverage is uneven. Major markets and large companies are covered in far more depth than small or illiquid ones.
No performance claims. We do not publish returns, win rates or back-tests, for TRUE or against anyone else. Such figures are trivially cherry-picked, are rarely independently audited, and tell you almost nothing about future outcomes. We'd encourage you to distrust them wherever you see them.
Frequently asked questions
Does TRUE make predictions?
No. It explains what has happened and what is currently driving an asset. It does not forecast prices, issue targets, or tell you what will happen next.
Can the AI be wrong?
Yes — which is exactly why every claim carries a source. Grounding and citation reduce error substantially, but you should verify anything that matters. That's the whole design: check, don't trust.
Why won't you publish a track record?
Because TRUE makes no calls to track. It doesn't recommend positions, so there's no performance to report — and we think published track records in this industry are generally more marketing than evidence.
Check our work.
Ask a question you can verify, and follow the sources.
For research and education. Not financial advice.