# A.I. Agents - Database

## Database Settings

This Agent skill allows searching information in files, spreadsheets, websites and text entered directly, aiming to improve its performance in its actions and in conversation with the lead.

<figure><img src="/files/a6c36d8b1bb91764b5c5b4295ac778e9e71ccb57" alt=""><figcaption></figcaption></figure>

### Import Format for Questions and Answers

The spreadsheet must contain two columns: one for the questions and another for the respective answers. The question-and-answer pairs must be entered starting from the second row, since the first row will be ignored.

After adjusting the format of your spreadsheet, click the 'Import CSV' button if the file is CSV, or 'Import XLSX' if it is XLSX.

<figure><img src="/files/cd1a6f1324f333cb74dc31c2ab28e9c2fcca3b48" alt=""><figcaption></figcaption></figure>

<figure><img src="/files/14a3a93b79d20a60602e397db0296a3984575e2d" alt=""><figcaption><p><em>The questions and answers spreadsheet format.</em></p></figcaption></figure>

### Website Extraction

To extract a website, you must first perform the mapping. For this, you should provide the URL or the route from which the mapping should be done, either at the root or, for example, at **/products**, checking the 'Extract only the above URL?' option. Then just click the 'Map' button.

<figure><img src="/files/00b0e5aa079174d278783ca156b1bd39c4c66eed" alt=""><figcaption></figcaption></figure>

### Selecting Extraction Method

The extraction can be done either with SprintHub's native tool or via the Firecrawl platform. Currently, [Firecrawl offers a free plan with 500 credits](https://www.firecrawl.dev/pricing), allowing extraction of up to 500 pages. It is recommended for more precise extractions and on sites with protection against crawling bots.

<figure><img src="/files/41bfaface2889f835386db7a13755010dd3f755e" alt=""><figcaption></figcaption></figure>

### Dynamic Extraction or Recurring Extraction

To keep the site's data always up to date, you can configure the 'Dynamic Extraction' option. In this mode, up to 20 pages from the extraction will be updated daily. To optimize the process, it is recommended to define a URL path that contains only the information that changes frequently. In addition, the 'Do not extract URLs containing' feature can be used to exclude certain paths or keywords from the extraction.

<figure><img src="/files/c41f4acf2edc68732e0a2ea2c82ad5d13d84c3ad" alt=""><figcaption></figcaption></figure>

### Link: How to Use Restriction by Tag/Class/ID

{% content-ref url="/pages/4a10fb09c6154a15fb2816611c91e9135a095d1f" %}
[AI Agents - How to Use Restriction by Tag/Class/ID](/en/topics/artificial-intelligence/ai-agents/ai-agents-how-to-use-restriction-by-tag-class-id.md)
{% endcontent-ref %}

### Files

When importing files, they are automatically transcribed, and just like website pages, it is possible to view and edit the text resulting from the extraction.

**Note**: To extract files you need to have a Gemini API key.

**Accepted file formats**: pdf, mpeg, mp3, wav, jpeg, mp4, mpg, avi, text/plain, csv, json, png (up to 50 MB per file).

<figure><img src="/files/4d42e906ce9bc3fb08d62fbf31e079b6b36a4130" alt=""><figcaption></figcaption></figure>

<figure><img src="/files/52f715ba9bed63aa99d1365a2ca0b6cadd72c461" alt=""><figcaption><p><em>Edit File Screen (extracted).</em></p></figcaption></figure>

### Text

In the field below, information can be entered for the agent to consult during the conversation with the lead. This field should contain only informational data, without rules, orders or behavior restrictions (these should be provided in the **Contextualization**).

<figure><img src="/files/53d5d2015d30cd8101aac04c7124bce955e4d841" alt=""><figcaption></figcaption></figure>


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.sprinthub.com/en/topics/artificial-intelligence/ai-agents/ai-agents-database.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.