How InvestIQ works

Editorial policy

Last updated: 6 November 2026

Key points
  • The AI explains figures computed by code; it does not produce them.
  • Texts are checked automatically against the computed figures and the evidence they cite. No person reviews each text before you see it.
  • No text is meant to tell you to buy, sell or hold; Deep Research reports are filtered for it.
  • AI outputs in the app carry a notice that they are educational guidance, not investment advice.
  • Errors can be reported to contact@getinvestiq.com; confirmed errors are corrected.

InvestIQ uses AI models to write: readings of an analysis, Copilot replies, Deep Research reports, War Room debates, audit summaries and newsletters. This policy sets the rules those texts follow and explains how they are checked.

1The role of AI

Figures come from the data and from InvestIQ’s code (see the methodology and the data sources). AI models read those results and turn them into text: what drives a score, what has changed, the bull and the bear case, what a filing says. They are instructed not to compute numbers and to rely only on the material they are given.

Stock Analysis itself — the score and its figures — involves no AI. The AI-written readings run only when you ask for them.

2How texts are checked

  • Copilot data cards: every figure on a card must appear in the data retrieved for your question. A card that carries a figure from the web, or that is worded as advice, is refused.
  • War Room: each figure an analyst quotes is checked by code against the evidence it cites.
  • Research modules: every piece of evidence a module cites must exist in the data it was given.
  • Deep Research reports: the writer cannot type figures; it places markers that code fills with the computed values. A compliance filter blocks ratings, buy, sell or hold verdicts, price targets and trading instructions, and a blocked report is not produced.

These checks are automatic. Free text — a Copilot answer, a sentence in a research module — can still contain an error they do not catch, and no person reviews each text before you see it. Check any figure that matters to you against its source.

3What the texts say — and do not say

The texts describe data: “the score is in the top 10% of the universe”, “the data indicates…”. The models are instructed never to say what you should buy, sell or hold, never to set a price target and never to size a position for you.

Opposing views are part of the method: the War Room sets a bull, a bear, a quant and a risk analyst against each other, and Stock Analysis shows the bear case next to the bull case.

AI outputs in the app carry a notice that they are educational guidance, not investment advice. See the disclaimer.

4Sources and citations

Deep Research reports link each claim to its source. War Room analysts cite the evidence they rely on, and research modules cite the data points they use.

5Models

The models in use, and the choice offered to you, are listed in data sources. InvestIQ picks a model per task — speed, depth, cost — and may change it; the rules above apply whatever the model.

6Articles on our blog

Blog articles are written by the InvestIQ team, with the help of AI tools, and signed “InvestIQ Team”. Each article shows its publication date and, once revised, the date of its latest update.

7Corrections

To report an error, write to contact@getinvestiq.com with the page, the stock and what looks wrong. We check it against the data and the filings. Confirmed errors in the data or the code are corrected; when an error comes from the method, the methodology and its date are updated.

A question about this document?
We answer every request, usually within a few days — contact@getinvestiq.com
Contact us