> ## Documentation Index
> Fetch the complete documentation index at: https://docs.baselayer.id/llms.txt
> Use this file to discover all available pages before exploring further.

# ask_question

> Ask a natural language question and get a synthesized answer.

# ask\_question

The primary tool for getting answers from your memory. It searches entities and facts, assembles context, and uses an LLM to synthesize a prose answer with citations.

This is the tool AIs should reach for first when answering questions about you, your projects, preferences, and history.

## Parameters

| Parameter  | Type   | Required | Description                                                  |
| ---------- | ------ | -------- | ------------------------------------------------------------ |
| `question` | string | Yes      | Natural language question to answer from the knowledge graph |

<Note>
  On the Free plan, `ask_question` is limited to 10 calls per month. Upgrade to Pro for unlimited queries.
</Note>

## Read-only

Yes. This tool only reads from your memory.

## Example usage

```json theme={null}
{
  "question": "What tech stack does the user prefer?"
}
```

```json theme={null}
{
  "question": "Who is their manager?"
}
```

```json theme={null}
{
  "question": "What hotel do they like when traveling?"
}
```

The response includes a synthesized prose answer with citations pointing back to the entities and facts that informed it.

## When to use

* Factual questions about the user, their work, or their preferences
* Questions about past decisions or conversations
* Any query where you need a direct, synthesized answer rather than raw search results

<Tip>
  `ask_question` is almost always the right starting point. If you need more detail after getting an answer, follow up with `get_entity` on the cited entities.
</Tip>

## When not to use

* Browsing or exploring what's in your memory -- use `memory_search` instead
* Looking for exact quotes from a past conversation -- use `retrieve_evidence`
* Writing new information to your memory -- use `record_memory` or `vault_fact`