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

# Ask Nucleus questions

> Learn how to phrase prompts and what project context Nucleus can see so you get accurate, useful answers during an audit.

Nucleus responds to natural-language prompts typed into the chat input at the bottom of the right-hand panel. The more specific your question, the more useful the response.

## What Nucleus can see

When you ask a question, Nucleus has access to the following data from the **currently open project**:

* All interview responses and notes captured so far
* AI readiness scores and dimension ratings
* Project metadata (client name, industry, stage, assigned team members)
* Any findings or recommendations already recorded

Nucleus does **not** have access to other projects, your organization's account settings, billing information, or external systems.

## How to phrase prompts

### Be specific about scope

Vague questions return vague answers. Ground your prompt in a specific area of the audit.

| Less useful              | More useful                                                                |
| ------------------------ | -------------------------------------------------------------------------- |
| "What's missing?"        | "Which AI readiness dimensions have no responses yet?"                     |
| "Summarize this"         | "Summarize the data governance findings in two sentences"                  |
| "What should I do next?" | "What follow-up questions should I ask about the client's MLOps maturity?" |

### Reference the audit stage

If you are mid-audit, tell Nucleus where you are:

> "We just finished the leadership interviews. What themes stand out so far?"

### Ask for structured output

Nucleus can format responses as lists, tables, or short paragraphs. Ask explicitly if you have a preference:

> "List the top three risks identified so far as bullet points."

## Example questions during an audit

**During discovery:**

* "What gaps exist in the client's current data infrastructure based on responses so far?"
* "Which stakeholders mentioned AI concerns? Summarize their key objections."

**During scoring:**

* "Which dimension has the lowest score and what drove that rating?"
* "Are there any dimensions where the evidence is thin or contradictory?"

**During reporting:**

* "Draft a two-sentence executive summary of the audit findings."
* "What recommendations should I prioritize for a client at this maturity level?"

## Limitations

* Nucleus can only answer questions about the **current project**. Switch projects to ask about a different client.
* Responses are based on the data entered into Audity, not on external research or real-time information.
* Long or complex prompts may result in truncated answers. Break them into smaller questions if needed.

## Next steps

* [Memory and context](/guide/nucleus/memory-and-context), understand what persists across your session and what resets.
