> For the complete documentation index, see [llms.txt](https://doc.aissist.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.aissist.io/tutorial/tune-aissist-behavior/refine-assets.md).

# Refine Assets

*Last updated: June 6, 2026*

Refine assets when Aissist gives incomplete, outdated, or inconsistent answers.

Update the source content, then test retrieval again.

See [Turn assets into AI](/tutorial/turn-assets-into-ai.md) for asset setup.

### Why refine assets

Assets shape the knowledge Aissist can retrieve in a conversation.

Refine them when you notice:

* missing operational details
* outdated policies or product facts
* repeated questions with weak answers
* conflicting guidance across sources

### What to update

Improve the source content, not just the reply.

Add the exact details users need to complete the workflow, such as:

* ordering steps
* policy exceptions
* required fields or documents
* links to the right page or form

### Reduce retrieval noise with sub agents

Associate assets with the correct [sub agents](/tutorial/create-sub-agents.md) to keep retrieval focused.

This reduces noise in the context Aissist uses for each reply.

If too many unrelated asset chunks are retrieved, answer accuracy can drop.

This can happen with any asset source, including websites, web pages, and Google Docs.

Use this pattern:

* link return assets to a return sub agent
* link order tracking assets to an order tracking sub agent
* link product policy assets to the sub agents that actually need them

### Check the response context in Session Detail

Use the session detail page to inspect how Aissist formed a reply.

For each AI response, review the context used to generate that response.

Look for:

* too many asset results in one reply
* asset content that does not match the current user question
* website or document chunks from unrelated workflows

If you see unrelated context, move those assets to the correct sub agent instead of leaving them broadly available.

This helps Aissist retrieve only the content that fits the active workflow.

### Example: Add missing purchase details

Open Goaaal often receives questions about replacement nets.

At first, Aissist gave a general answer. It confirmed replacement nets were available, but it did not explain how to place the order correctly.

The team fixed this by updating a Google Docs asset with the missing workflow details:

> **User asks about replacement nets**
>
> To order a replacement net:
>
> 1. Go to `https://opengoaaalusa.com/products/spare-parts`
> 2. Select the spare replacement net option.
> 3. Choose the quantity.
> 4. In the message field, include the product size or model and confirm it is for a replacement net.
> 5. Complete checkout.
> 6. Email the order number so the team can expedite dispatch.

After that update, Aissist returned clearer and more actionable answers.

The improvement came from refining the asset, not from adding a broader instruction.

### Best practices

* update the source where the missing detail belongs
* associate assets with the right sub agents to keep retrieval focused
* use session detail to spot unrelated asset context in AI replies
* keep operational steps explicit and easy to retrieve
* test asset retrieval after each major change

{% hint style="info" %}
Use the [Asset Debugger](/tutorial/turn-assets-into-ai/asset-debugger.md) to check whether Aissist can retrieve the exact content you expect.
{% endhint %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## 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, and the optional `goal` query parameter:

```
GET https://doc.aissist.io/tutorial/tune-aissist-behavior/refine-assets.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

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.