> ## Knowledge Base Index > Fetch the complete knowledge base index at: https://help.crisp.chat/sitemap.xml > Use this file to discover available pages before exploring further. > Pure-Markdown content can be obtained by appending a '.md' suffix to the content URLs listed in the sitemap (without the trailing slash). # How to use Hugo AI Suggestions to improve Customer Support knowledge *This article explains how to use Hugo AI Suggestions to identify missing, outdated, or incomplete support knowledge based on real customer conversations.* [Hugo AI Agent](https://crisp.chat/en/ai/) uses your training resources and guidance to answer customer questions. As your product and customer needs evolve, some information may become incomplete, outdated, or missing entirely. AI Suggestions analyzes customer conversations and turns recurring knowledge gaps into concrete additions, edits, and removals that you can review before updating your content. ||| **Hugo AI Suggestions is a premium feature available exclusively on the Crisp Plus plan.** If your workspace is on [another plan](https://crisp.chat/en/pricing/#enterprise), you will need to upgrade to Plus to access this feature. __In this guide, you will learn how to:__ * [Understand Hugo AI Suggestions](#1-what-is-hugo-ai-suggestions) → see how conversations are turned into knowledge improvements * [Know when to use AI Suggestions](#1-when-should-you-use-ai-suggestions) → identify missing or outdated support information * [Open AI Suggestions and understand the analysis](#1-open-ai-suggestions-and-understand-the-analysis) → review the AI digest, themes, and generated suggestions * [Review suggested improvements](#1-review-suggested-improvements) → understand what Hugo recommends changing and why * [Review and apply a suggested change](#1-review-and-apply-a-suggested-change) → approve, reject, and save proposed modifications * [Manage your suggestions](#1-manage-your-suggestions) → search, filter, configure, and re-analyze suggestions * [Understand AI Suggestions vs Analytics vs Observability](#1-ai-suggestions-vs-analytics-vs-observability) → choose the right tool for your analysis * [Follow AI Suggestions best practices](#1-best-practices) → review recommendations before updating your knowledge --- # ${color}[#0080dd](What is Hugo AI Suggestions) Hugo AI Suggestions analyzes customer conversations to identify information that could be added, updated, or removed from the knowledge used by your AI Agent. Instead of manually reviewing conversations one by one, the feature identifies recurring themes and generates concrete improvements based on what customers actually asked. __Depending on what it detects, AI Suggestions can recommend:__ * Adding information that customers repeatedly ask about * Updating existing content when important details are missing * Correcting information that no longer matches recent conversations * Removing content that appears outdated * Improving existing Articles or Snippets with additional context Each suggestion is connected to the conversations that contributed to it, helping you understand why the recommendation was generated. || AI Suggestions does not automatically rewrite your content. Proposed changes must be reviewed and approved before they are saved. To understand the other areas available in Hugo, read [**How to navigate the Hugo interface and features**](https://help.crisp.chat/en/article/how-to-navigate-the-hugo-interface-and-features-h1eatv/). --- # ${color}[#0080dd](When should you use AI Suggestions) AI Suggestions is useful when you want to understand whether your current support knowledge still matches what customers are asking. #### ${color}[#445055](Identify missing knowledge) If several customers ask about the same topic and the necessary information is missing from your content, AI Suggestions can surface that recurring gap. You can then review the proposed addition and decide whether it should become part of Hugo's knowledge. #### ${color}[#445055](Keep existing content up to date) Customer conversations can reveal that an existing Article or answer is incomplete or no longer accurate. AI Suggestions can identify the relevant source and propose a specific change instead of requiring you to find it manually. #### ${color}[#445055](Turn recurring support questions into improvements) AI Suggestions connects what customers actually ask your support team with the knowledge available to Hugo. This can help you identify recurring questions that require better coverage and reduce gaps between your current documentation and real customer needs. To manage the information available to Hugo directly, read [**How to train Hugo AI Agent on your data**](https://help.crisp.chat/en/article/how-to-train-hugo-ai-agent-on-your-data-ijkdrl/). --- # ${color}[#0080dd](Open AI Suggestions and understand the analysis) To access AI Suggestions: 1. Open [**Crisp**](https://app.crisp.chat) 2. Select **AI Agent** 3. Go to **Evaluate → AI Suggestions** ![](https://storage.crisp.chat/users/helpdesk/website/-/8/7/a/e/87ae2703583ac800/hugo-ai-suggestions-analysis-r_c3a1q5.png =1000x566) The page is divided into two main areas: * **Analysis recap** → understand what customers talked about during the analyzed period * **Suggested improvements** → review the concrete content changes generated from those conversations #### ${color}[#445055](Understand the Analysis recap) At the top of the page, the **Analysis recap** summarizes the latest conversation analysis. You can use the recap selector in the top-right corner to view the available analysis period. __The recap includes:__ * **AI digest** → a summary of the most notable themes detected across customer conversations * **Conversations** → the number of conversations included in the analysis * **Themes** → the number of recurring subjects identified * **Suggestions** → the number of improvements generated from the analysis Expand **What customers wrote about** to review the individual themes detected across the analyzed conversations. For each theme, you can see examples of what customers discussed and whether an AI suggestion was generated from that topic. This gives you context before reviewing the recommended content changes below. --- # ${color}[#0080dd](Review suggested improvements) The **Suggested improvements** section contains the recommendations generated from the conversation analysis. Suggestions can propose changes to content such as **Articles** or **Snippets**. __Each suggestion can include:__ * The content type concerned * The recommended addition, edit, or removal * A short explanation of the issue detected * The number of conversations supporting the recommendation * An indication of the strength of the signal For example, several customer conversations may reveal that an existing help article is missing an important setup step. AI Suggestions can identify that article and propose adding the missing information. Use the supporting conversations and context to understand whether the recommendation represents a recurring customer support issue before applying it. || A higher conversation count can indicate a recurring knowledge gap, but it does not automatically mean that the proposed change should be accepted. --- # ${color}[#0080dd](Review and apply a suggested change) Open a suggestion to inspect the exact modification before applying it. Depending on the source, AI Suggestions can display the existing content alongside proposed additions, removals, or replacements. ![](https://storage.crisp.chat/users/helpdesk/website/-/8/7/a/e/87ae2703583ac800/review-and-approve-hugo-ai-sug_4qoev2.png =900x654) __When reviewing a suggestion, you can:__ * **Accept** a proposed modification * **Reject** a modification you do not want to apply * Use **Accept All** or **Reject All** when several changes are proposed * Review how many changes remain pending, accepted, or rejected * Click **Save** once you are satisfied with the final result Suggested additions and removals are highlighted directly inside the content so you can understand exactly what would change. For longer Articles, a suggestion can contain several separate modifications that can be reviewed independently. ||| Always review the proposed content before saving it. AI Suggestions uses customer conversation evidence to recommend improvements, but you remain responsible for validating the final information. --- # ${color}[#0080dd](Manage your suggestions) The Suggested improvements area includes several tools to organize recommendations and control how new ones are generated. #### ${color}[#445055](Search and filter suggestions) Use the controls above the suggestion list to narrow what you want to review: * **Search** → find a specific suggestion or topic * **Action** → filter recommendations by the type of change proposed * **Content type** → focus on specific sources such as Articles or Snippets This is useful when your workspace contains a large number of suggestions. #### ${color}[#445055](Understand suggestion statuses) Suggestions are organized into four groups: * **Improvements** → suggestions currently available for review * **Outdated** → suggestions whose underlying source changed after the recommendation was generated * **Approved** → suggestions that have already been accepted * **Ignored** → suggestions that have been dismissed The number next to each group shows how many suggestions currently belong to that status. The **Outdated** status is particularly important. When the source changes after a suggestion was generated, the original recommendation may no longer match its latest version. Review the current source before making another change. #### ${color}[#445055](Configure suggestion settings) Click **Settings** from the Suggested improvements section to configure how AI Suggestions generates recommendations. ![](https://storage.crisp.chat/users/helpdesk/website/-/8/7/a/e/87ae2703583ac800/hugo-ai-suggestions-settings-m_ss21qs.png =700x517) __You can customize:__ * **AI Model** → choose the model used to analyze conversations and generate suggestions * **Output Language** → choose the language used for generated content * **Custom Instructions** → provide writing style, tone, formatting, or other guidelines that suggestions should follow Custom Instructions can help generated changes better match your existing documentation standards before you review them. Click **Save Settings** to apply the configuration to future suggestions. #### ${color}[#445055](Re-analyze your conversations) Click **Re-analyze** when you want AI Suggestions to generate a fresh analysis based on your available customer conversations. This can be useful after making meaningful changes to your support knowledge or when you want to review more recent customer questions. The **Ignore current page** action is also available from the Suggested improvements view when you want to dismiss the suggestions currently displayed. --- # ${color}[#0080dd](AI Suggestions vs Analytics vs Observability) AI Suggestions, Analytics, and Observability all help you improve Hugo, but they answer different questions. | Feature | Best used for | | ---- | | **AI Suggestions** | Identifying concrete knowledge improvements from multiple customer conversations | | **Analytics** | Measuring overall AI Agent performance, activity, outcomes, and conversation trends | | **Observability** | Understanding and debugging what happened inside one specific AI Agent conversation | For example: * Use **Analytics** to identify broader customer support trends * Use **AI Suggestions** to turn recurring questions into concrete content improvements * Use **Observability** to understand why Hugo handled, escalated, or failed one specific conversation To explore these tools in more detail, read [**How to use Hugo Analytics to improve your AI Agent**](https://help.crisp.chat/en/article/how-to-use-hugo-ai-analytics-6ymdxs/) and [**How to monitor and debug Hugo AI Agent conversations with Observability**](https://help.crisp.chat/en/article/how-to-monitor-and-debug-hugo-ai-agent-conversations-with-observability-1q79tp1/). --- # ${color}[#0080dd](Best practices) AI Suggestions is most useful when recommendations are treated as evidence-backed improvements rather than automatic changes. #### ${color}[#445055](Review the conversation evidence) Check the customer conversations and themes behind a suggestion before deciding whether the proposed change is relevant. #### ${color}[#445055](Prioritize recurring gaps) Start with suggestions connected to questions that repeatedly appear in customer support conversations. Improving these areas can give Hugo better information for common customer requests. #### ${color}[#445055](Review outdated suggestions carefully) When a recommendation becomes **Outdated**, check the latest version of the underlying source before making additional changes. The original issue may already have been addressed. #### ${color}[#445055](Use the right place for each type of information) Factual support information should remain inside the appropriate Hugo training resource. Use [**Hugo Instructions**](https://help.crisp.chat/en/article/how-to-write-hugo-instructions-for-ai-agent-and-copilot-1qc131v/) for behavior and response rules, and [**Hugo training resources**](https://help.crisp.chat/en/article/how-to-train-hugo-ai-agent-on-your-data-ijkdrl/) for factual support knowledge. #### ${color}[#445055](Re-analyze after meaningful updates) After improving several recurring knowledge gaps, run a new analysis later to identify what customers are asking about next. This helps keep the knowledge available to [**Hugo AI Agent**](https://crisp.chat/en/ai/) aligned with evolving customer support needs.