# How Areinum finds and checks information behind an answer

Suppose you ask, "What changed in this project?" A useful answer needs more than a plausible summary. It needs the right records, their dates, and a way to show you where the important details came from.
Areinum uses a persistent graph to help find related records and tools to perform defined operations. You can explore this flow in the pre-production demo with fictional company data.

Canonical: https://areinum.com/guides/how-areinum-finds-and-checks-information/
Language: en
Updated: 2026-10-31

## From a question to relevant records

A request identifies a matter, a customer, a person, a date, or another useful clue. In the demo, the graph indexes metadata such as titles, names, dates, and identifiers, helping the assistant find related records.

The graph indexes metadata, not the text of messages or attachments. When more detail is needed, the assistant can read selected record text. The scope and freshness of the available sources still affect the answer.

1. Define the matter and the period you are asking about.
2. Find the relevant records and identify their sources.
3. Read the selected details needed for the question.
4. Check the result and any missing or conflicting information.


## An example: a changed project scope

Imagine a fictional contractor asks what changed in a deck project. An original scope lists the deck surface; a later customer message asks for new lighting. The useful result names the addition, cites the two records, and states whether approval of the change is visible.

If the available material includes a proposed price but no acceptance, the answer should say that the price is proposed and identify what still needs the customer's decision.


## What Rust and the tools contribute

The business tools were rewritten in Rust and can be tested in the platform. They include tools for file work and calculations, alongside graph operations for finding, counting, grouping, and summing relevant records.

Code handles defined operations while the assistant brings the result into the conversation. The time needed depends on the task and the information involved.


## Read the answer with the right questions in mind

Check that the answer covers the requested period and the right customer or job. Inspect the source behind any consequential amount or statement, and compare the date of the record with the data-up-to information when it is provided.

If a source is inaccessible or two records disagree, resolve that gap before relying on the answer. A missing result may reflect incomplete information; check another appropriate source.


## Questions you may have

### Does the graph contain every word in every file?

The graph indexes metadata, not the text of messages or attachments. The assistant can read selected record text when needed. Check which files and records are available for your task.

### Why use code for a calculation?

A defined calculation can be performed by a tool using the supplied values rather than relying on a conversational answer alone. The inputs, units, scope, and result still need to be checked.

### Does Rust guarantee a response time?

No. Rust is used for the business tools, but response time depends on the task and the information involved.

### What if I cannot find the source behind an answer?

Treat the statement as needing verification. Ask for the source if it is available, or check the original record through an appropriate route before relying on the answer.

## Related reading

- [Find the details behind a business question](https://areinum.com/product/company-knowledge/)
- [Find the information behind an answer](https://areinum.com/product/data-checks/)
- [How to check work prepared by AI](https://areinum.com/guides/check-work-prepared-by-ai/)

Explore the demo: https://platform.areinum.com/

The demo uses sample business data and is in English. Access may need to be arranged.