Optimizing your help center content for AI agents

What's my plan?

Suite

Any plan

If you're using features such as AI agents that draw information from your help center content, focused, well-written, and well-formatted articles will lead to better results. You can optimize your articles to improve the quality and accuracy of automated responses that include generative replies and suggested articles.

Note: In some older accounts, autoreplies are called article recommendations.

This article contains the following topics:

• About automated replies based on help center content
• Planning and organizing article content
• Using article labels to filter results (autoreply with articles only)

Related articles:

• Best practices: Preparing your help center for generative AI

About automated replies based on help center content

Some features can assess your help center content to send relevant, helpful responses to an end user’s request sent via email, web form, or messaging. These responses can be sent as:

• Links to suggested articles. A list of links for up to three specific articles. This type of reply is used in autoreplies with articles for web forms and emails.

• Generated responses. AI-created replies sent in response to an end user’s free-text entry in an AI agent conversation. This type of reply is used in generative replies in an AI agent.

These actions may be counted as an automated resolution. See About automated resolutions for more information.

Planning and organizing article content

The way you organize the content of a help center article impacts how effectively an AI agent can locate relevant information.

Creating titles for your articles

Article titles play a big part in determining whether an article is a good candidate for solving a customer's problem. If an article's title closely matches a customer’s text in an email or web form support request, or free-text comment in an AI agent, it's more likely to come up as a suggestion in the automated response.

Give your articles titles that are similar to how your customer might phrase their support request. Try formatting your article titles in one of the following ways:

• As a question : "How do I reset my password?"
• As a simple, active phrase: "Resetting a password"

Narrowing an article's focus

When possible, an article should cover a single, specific topic. How you implement this suggestion depends on how you present information in your articles:

• Issue/resolution articles: Articles that address specific issues and resolutions should focus on a very specific issue with a single resolution. For instance, rather than having a single article covering multiple related billing issues – payment options, requesting a refund, and challenging a charge – write three articles, one for each billing issue. Offer a single resolution for each issue to guide your customers to your preferred resolution, and link to other articles that may offer alternative resolutions.

• Process articles: The same principle applies to articles that describe processes and procedures. For example, when documenting an account set-up process that involves two distinct tasks (creating an account and selecting account options), consider presenting the information in two separate, concise articles, one for each task. This helps the AI agent identify the article that best matches a customer's specific request.

Using article labels to filter results (autoreply with articles only)

Labels can restrict the articles are considered in an autoreply with articles response. After you add a label to an article, you can include it in the autoreply with articles trigger to filter the results.

There are a few approaches you can take when using labels.

Reducing the “noise” in your help center

Your help center may include articles that should not be available for automated article retrieval results. In this case, add a "use_for_autoreply" label to the articles you want the AI agent to consider. Add this label to all autoreply with articles triggers.

Note: This approach can only be applied to autoreplies with articles for email responses.

Targeting customer segments

You can use article labels in triggers when you have different customer segments and you want to show each segment only the relevant articles. For example, suppose you are a mobile game developer and you support both Android and iOS platforms. When you get a request from a customer who's using Android, you want to suggest only Android-related articles.

To accomplish this, create an "Android" autoreply trigger with the condition based on your custom field "Platform = Android." Then, configure the trigger using labels to include only articles that contain the "android" label.

Repeat the process to set up a trigger for the iOS platform and label.