> ## Documentation index
> Fetch the complete documentation index at: https://docs.prescientai.com/llms.txt
> Use this file to discover all available pages before exploring further. Add `.md` to any page URL to get that page as Markdown.

# Ask a question

> How to send a question to Prescient AI, follow the answer while it is prepared, read charts and tables, and open the sources of an answer.

You ask Prescient AI a question in plain language. Prescient AI reads the data of your workspace and answers in the chat.

## How do you ask a question?

1. Open **Prescient AI** from the sidebar.

2. Type your question in the message box, or tap a reply pill.

   A reply pill sends a question as your message. The pill with the label **Last week** sends a question about the cause of the change that it names, not the text of its label.

3. Send the message.

   The send button becomes a stop button while the answer is in progress. Select it to stop the answer.

4. Read the answer.

   To continue, type a follow-up question. Prescient AI uses the earlier messages of the chat as context.

After your first message, the address of the page changes to the address of the chat. The chat appears in **History** after the next load of the page.

> **Name the period:** An answer covers the last 90 days unless your question names a period. To get a different period, put the dates in the question. One answer can cover up to about 2 years.

## What happens while the answer is prepared?

The chat shows what Prescient AI is doing. It shows **Thinking…** first. When Prescient AI reads your data, a line names each step, for example **Checking your attribution…**. For each status, see [Chat reference](https://docs.prescientai.com/reference/prescient-ai/chat#status-lines).

A question that needs campaign data takes longer. After 12 seconds with no new output, the chat says so.

## How do you read an answer?

An answer can contain several parts.

- **Text.** The answer in words, with tables where a table is clearer.
- **A visual.** A chart, a table with pages, or a view from Attribution or Media Forecaster. See [Visuals in an answer](https://docs.prescientai.com/reference/prescient-ai/visuals).
- **Sources.** When the answer uses the knowledge base, the articles are listed under **Sources**. Select an article to read it in a dialog.

Figures in an answer come from the same data as the [Attribution](https://docs.prescientai.com/attribution), [Model Center](https://docs.prescientai.com/model-center), and [Media Forecaster](https://docs.prescientai.com/media-forecaster) pages.

## How do you choose the model that answers?

The message box has a model picker. The first option is the default and suits most questions. The chat keeps your choice for its later messages.

## What if an answer fails?

| Message                                     | What to do                                                                                                     |
| ------------------------------------------- | -------------------------------------------------------------------------------------------------------------- |
| **Something went wrong. Please try again.** | Send the message again. This message also shows when a message is too long or when you send messages too fast. |
| **This answer didn’t finish. Ask again.**   | The answer reached the time limit. Send the question again, or ask a narrower question.                        |

For each limit of a chat, see [Chat reference](https://docs.prescientai.com/reference/prescient-ai/chat#limits).

## What can a chat not do?

- A message contains text only. You cannot attach a file or an image.
- You cannot send a message while **Build this plan** prepares a plan card. The message box then reads **Creating your plan…**.
- A chat has no control to rename it, delete it, copy an answer, or rate an answer.
- A chat is private to you. To share a conversation, [save it to the Library](https://docs.prescientai.com/library/save-a-conversation).