> For the complete documentation index, see [llms.txt](https://matterhorn-doc.mometic.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://matterhorn-doc.mometic.com/investigate-the-business/ask-matterhorn.md).

# Ask Matterhorn

Use Ask Matterhorn to retrieve company records, understand grounded answers, and turn a broad investing question into a focused investigation.

**Ask Matterhorn** helps you locate and summarize records already in the platform. It sits above **Top 10 Opportunities** and starts collapsed on each page load so Overview stays focused on your research list.

## Ask a question

1. On **Overview**, select **Show** beside Ask Matterhorn.
2. Choose a suggested question or type your own.
3. Press **Enter** or the arrow button to submit. **Shift+Enter** adds a line break.
4. Read the answer, retrieved-record count, and any warnings.
5. Open linked tickers to examine the underlying company records.

![Ask Matterhorn expanded with a compounding-rate question ready to submit](https://981865776-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FNOJXLd7UJBniObM136Nu%2Fuploads%2Fgit-blob-f3db981e76c39b115e4d1ba986df25b9d1b6a925%2Fask-matterhorn.png?alt=media)

*This screenshot shows a question ready to submit, not a generated answer.*

## Start with questions the records can answer

* “Which companies have the highest compounding rate?”
* “Show the strongest forward-demand companies.”
* “What does Matterhorn have on NVDA?”
* Use the live suggested questions when you are unsure which retrieval paths are supported.

The retriever recognizes supported question patterns. It is not an unrestricted query language or a guarantee that every condition in a complicated prompt will be applied. Inspect the returned companies, intent label, and records to confirm that the answer addresses your question. For precise combinations of filters, use **Companies**.

## What the writer can and cannot do

The database supplies the records; the configured language model writes the explanation from that evidence. For a question about why a company stands out, check the supporting facts as well as the score. Without a configured model, Ask can still return the retrieved records directly. The analyst reviews use the same configured service, although their dossiers and tasks differ.

Ask is not a live web-research engine. It cannot guarantee that an earnings release published moments ago has entered the evidence record. The ML toggle also does not turn every Ask retrieval into an ML-ranking query. Check the returned basis.

> **Pro Tip — Ask for the evidence you need next.** Use the answer to reach a source or a company, then inspect it. A smooth explanation is useful navigation, not an additional independent source.

Related: [company filters](/find-opportunities/rankings-and-filters.md), [evidence verification](/investigate-the-business/evidence-and-citations.md), [freshness](/keep-research-current/refresh-and-freshness.md).