Home Content | Managing ClaudIA's Knowledge
📖

Content | Managing ClaudIA's Knowledge

How to use the Cloud Hub to improve content for increased quality and retention
Fabrício Rissetto Ian Kraskoff
By Eduardo Fukumoto and 4 others
• 9 articles

Using the content screen

Content The content screen displays all sections of IDS (Improved Data Set) Search and filter by sections On this screen, it is possible to sort the sections and apply filters such as: - Exact search (equivalent to CTRL + F in the browser) - Semantic search (an intelligent search using Vector DB technology to bring contents that most resemble the search terms) - Type: Sections N1, N2, or INTERACTIVE (Controlled Flow) - Tag - Flow (in case of using the Multiprompt feature - which is not applicable by default in all projects) Viewing the modification history of a section On this screen, you can see the modification history of each section by clicking on the "history" icon: Clicking the button will open a window with all modifications of that content since its creation date: When clicking on "View details," it shows exactly what was modified in each version: Creating a new section To create a new section, simply click the "Create New" button. It will open a second screen where you can enter the title of the section, the response of the section, and define whether it is a N1, N2, or Iterative section. Remember that an Iterative section is one that will trigger a Controlled Flow and the tag related to that section. Finally, just click the "Save" button. Editing an existing section To edit a section, simply click the edit button or double-click directly in the middle of it, and it will open the editing screen. On the editing screen, you can change everything: the title, the response, the type, and the tag. Once you're done, you can click the green checkmark (✅) at the top right if you want to save or the gray x (❌) on the right if you want to discard your changes. Temporarily disabling a section A relatively new feature is the ability to temporarily disable content, temporarily disabling a session from the IDS. Let's suppose there is a piece of content that, for the moment, does not make sense, but in the future it does. For example, you are about to launch a new product and want to create this content well before the rollout of this product. You can write this section and disable it by clicking the brown eye button (👁️) at the top right of the section. This will prevent it from being used by ClaudIA. When you want to enable it again, just click that button again. While it is disabled, it will be marked prominently with the word "Disabled" Deleting a section from the IDS If you want to delete a session from the IDS, just click the red trash icon (🗑️) at the top right of the screen. This will open a pop-up asking you to confirm that you really want to delete it. If you click confirm, it will be deleted. If you click cancel, it will not delete and will cancel the deletion you started.

Last updated on Jun 24, 2026

How to save IDS sections as drafts or schedule them for later publication

It has now become much easier to create and manage temporary content and drafts in the IDS. This feature was created because many clients wanted to prepare content in advance — whether to review before final validation or to have everything ready before changes to policies or products come into effect. Creating new content: draft or scheduling To get started, access the content section in the Hub. Image First, click on "Create New" and fill in the fields: - Title - Response - Type (N1, N2, or Interactive) - Tag - Flow (if Interactive) - Topic (if you use this function) Next, you can view the publication options by clicking the arrow to the right of the Publish button. At this moment, you can choose between: - Publish immediately - Save as draft - Schedule a date and time for publication Image Save as draft This option allows you to continue editing the content later. While it is saved as a draft, ClaudIA will not have access to it. Schedule publication When choosing to schedule, a window will open to select date and time. After clicking "Schedule", the content will be automatically published at the defined time! Image Important: the publication date is pre-filled. - If this is your first scheduling, the system will suggest the date of tomorrow at 9 AM. - If you have already scheduled another content recently, the system will use the last date and time used. Editing content You can also edit content that has already been created and: - Save the changes as a draft - Schedule for later publication - Publish immediately Once you make changes, the three options will appear for you, functioning the same way as when creating new content. Image These new functionalities of the IDS bring more flexibility, organization, and efficiency for those who need to plan, review, and manage content in advance.

Last updated on Aug 12, 2025

Best Practices in Creating and Maintaining Content

Keep the content base with less than 20% N2 sections Understand the composition of the content base The content base is composed of N1, N2, and Interactive sections (when Controlled Flows are being used). The ideal content base in terms of retention is one that contains only N1 and/or Interactive sections and does not escalate requests to humans, making ClaudIA more effective and providing faster responses to customers (compared to humans). Consider the quality of service However, the quality of service may be affected if we try to automate 100% of requests. There are situations where a human agent needs to intervene. For these cases, we create N2 sections. Avoid creating a queue of requests If all sections were N2, we would have no level of automated service. This would leave the retention level at 0% and could generate a queue of requests, also affecting the quality of service. Seek balance between N1 and N2 sections To maintain good levels of retention and quality, it is essential to balance the sections that automate service (N1) and those that escalate to human assistance (N2). Follow the recommended maximum percentage of N2 sections We recommend that the maximum percentage of N2 sections in a knowledge base be 20% of the total. For example, in a knowledge base with 1000 sections, ideally, up to 200 should be N2. Write the section title clearly and without spelling errors Imagine that the user sent a message that ClaudIA could not respond to correctly because there was no content in its base. You know the answer to the customer's question, which leads you to create a new section in the content base. In situations like this, or when there is a need to create or change some content, it is crucial that the texts are clear and free of spelling errors. This increases the likelihood of ClaudIA understanding the user's question and locating the correct section to respond. Tip: Do not insert the customer's message in the section exactly as they wrote it. Correct the spelling and, if necessary, make the question clearer. See an example of a section with an inadequate title: Now, observe an example of a session with an appropriate title: Remove unnecessary sections After identifying that a section is being frequently used incorrectly (in this video, we teach how to map these cases), delete it. In many cases, these sections were created deviating from best practices, and there is no problem in deleting them, especially if they are N2. Constant maintenance of content is a lever to drastically improve the quality and retention capacity of tickets by ClaudIA. Video on how to map retention offenders. Template to map N1 opportunities based on N2 sections and for ticket analysis (section was used/section was used correctly). Delete duplicate sections Duplicate Sections: What Are They and Why Avoid Them? Duplicate sections are those that have identical titles but present different answers or forwarding. In addition, sections with similar titles that should be activated in the same situations but contain distinct answers are also considered duplicates. Why avoid duplicate content? NEVER create duplicate content, especially with different types (N1, N2, or INTERACTIVE). This practice can cause various problems, such as: - Difficulty in content maintenance: repeated content generates constant rework. - Risk of incorrect answers: increases the chance of ClaudIA selecting the wrong section to respond to the user's query. Example of Duplicate Sections In the examples above, in addition to the sections having different responses, they also imply distinct directions: the first is an N1 case, while the second is N2. Note that in these cases, ClaudIA tends to prefer to escalate the ticket to N2. Example of the Impact of Duplicate Sections How to identify duplicate sections? There are two tools in the Hub that can be used to identify duplicate content: - Content screen, using the "semantic" filter: - Playground In both cases, the suggested step-by-step is: 1. Access the tool; 2. Add a question with possibly duplicate content; 3. On the content screen: check if there are two or more sections with similar titles - and, mainly, different responses; 4. In the Playground: check if the answer was correct and click on the "i" to check if another similar section was present in the topK; Both tools work like a semantic search mechanism, that is, they search for content similar to the added question. Do not include greetings in IDS sections Avoid using greetings like 'Hello! How are you?' along with responses in the sections. ClaudIA already greets the user at the beginning of the conversation or when necessary. Understand that this practice can intensify the feeling of "robotization" of service for the customer. It can also increase the chance of ClaudIA triggering an incorrect section when the user greets. This also harms the results of ClaudIA's tagging tool, the Tagger. See below an example of an inadequate section: Do not create "no content" sections in IDS Similar to the above item, avoid creating sections that do not have content. Like sections that try to adapt ClaudIA's tone of voice: - Title: "Okay, thank you" - Response: "You're welcome, I'm here to help you!" Sections like this hinder ClaudIA's performance, adding noise to its knowledge base.

Last updated on Jun 24, 2026

Content Segmentation by Topic

ClaudIA allows for the segmentation of content for different aspects of your business through the functionality of segmenting IDs by 'topics'. For example, we can have a specific topic for customers in the 'RETAIL' segment. Through a Controlled Flow, we can: - Automatically query a customer attribute in the helpdesk or back office, without the need to ask them directly (as demonstrated in the image below). - Ask the customer (if the information is not available automatically) about the segment they belong to or about the topic of interest. When the Controlled Flow detects that the customer belongs to a specific segment, the conversation will only use the content registered for that segment. In the example below, the response provided was based on the content of the 'RETAIL' topic: By auditing the response, we can see the topic of the section that was used: To filter, edit, or create new content in topics, the Content screen can be used: Use of the "DEFAULT" Topic The "DEFAULT" topic stores content common to all segments. For example, generic questions like: "What services does your company offer?". Scenario Examples - At the beginning of the conversation, before a topic is selected by the Controlled Flow, only the content from the "DEFAULT" topic is used. - After selecting a specific topic, such as "RETAIL", the system will only use the content from the "DEFAULT" and "RETAIL" topics to respond to the customer. This topic can be configured on the following Controlled Flow screen: Final Considerations Currently, for a topic to be added to the IDS sections, it is necessary to request a Clouder to register the topics in the database beforehand. In the future, we intend to provide a screen for clients to manage their own topics in a self-service manner.

Last updated on Jun 24, 2026

Synchronized FAQ or Product Page Content

How it works Starting from a URL, we can use our agent that extracts information from your site and prepares it for ClaudIA to use. The content is optimized so that Claudia can maintain performance. What can and should be avoided What can be done: - URLs that start with "help" or "support" - URLs that have the base website as the beginning www.example.com and contain a '/products' path that we want to extract - URLs that have the base website as the beginning www.example.com and include some '/blog' paths, for example, which we do not want to extract What to avoid: - URLs where all information is contained on a single page How you can view this in the Content page How to handle this content You will be able to view and access the External Content that was used as the basis for generating this specific session. Additionally, a filter can be applied to view all sources extracted from external origins. Finally, based on certain filters, it is possible to disable or delete multiple IDS sessions if desired. This external content will not be editable, as it will always be overwritten when performing another synchronization with the content database. Which tag gets filled Initially, a default arbitrary tag will be filled in (e.g., 'faq' or 'product') ⮕⮕⮕ We plan to evolve towards automatic content categorization in future versions. How you will see it in the Conversation Just like in the "Content" tab, in the Conversation tab you can also see whether the content is from an external source or not. How to perform synchronization IMPORTANT: If you choose to fetch external source content using our agent, it is ESSENTIAL to disable sessions that are identical or duplicated. Our team can analyze duplicates to support this process. ⮕⮕⮕ Currently, you need to contact our support team to use this feature; it is in Beta. 📌 List of HTML tags ignored by default During the refinement process, the extraction agent automatically discards many elements that are not usually useful for the knowledge base. These selectors are defined to reduce visual and structural noise (menus, ads, comments, forms, etc.), keeping only the main content. Page structure #footer #header #nav nav footer Scripts and styles script style noscript Media svg img audio video Navigation and menus .sidebar .menu .navigation .breadcrumb .breadcrumbs .pagination .pager .page-navigation Ads and banners .advertisement .ads .ad-banner .cookie-banner .cookie-notice .gdpr-notice Social and sharing .social-share .social-buttons .share-buttons Forms and subscriptions .newsletter .subscription .signup-form .search-box .search-form .search-bar Related content .related-posts .recommended .suggestions Comments and discussions .comments .comment-section .discussion Metadata and authorship .tags .tag-list .categories .author-bio .author-info .byline .meta .metadata .post-meta Widgets and sidebars .widget .widgets .sidebar-widget Popups and overlays .popup .modal .overlay .lightbox Accessibility and hidden navigation .skip-link .screen-reader-text .sr-only .print-only .no-print Accessibility Attributes [role='alert'] [role='banner'] [role='navigation'] [role='complementary'] [role='dialog'] [role='alertdialog'] [role="region"][aria-label*="skip" i] [aria-hidden='true'] [aria-modal='true'] Invisible elements .hidden .invisible

Last updated on Aug 25, 2025

Synchronize External Content for AI (External Source)

This FAQ explains in a practical way how to register, edit, activate/deactivate, and synchronize content sources of your project — as well as interpret statuses and configure collection recurrence. What are "Content Sources"? They are settings that specify where ClaudIA should search for knowledge (e.g., your Help Center). Each source has: Initial URL, path filters, scheduling/clock for synchronization, and a default tag. Where do I find the screen? In the project, go to Content → Sources. Below is an example of the screen in a project with no active synchronization yet. If the project already has active syncs, you'll see a list with the following columns: - Initial URL - Status of the last execution (e.g.: Synchronized, Error, Canceled) - Last synchronization - State (Activated/Deactivated) - Actions: Synchronize, Edit, View URLs You can sort by clicking on the column headers. How do I add or edit a source? 1. Click on Add Source (or Edit an existing one). 2. In the modal, fill out: - Scheduling (Recurrence & Local Time) Select Daily / Weekly / Monthly and a time. The interface displays: - The next execution (relative and with local date/time). - The system automatically converts everything to UTC cron. - Initial URL Must be a valid URL with protocol http/https. - Path Filter (mandatory) Synchronizes only pages whose path starts with one of these prefixes (one per line). → Each value must start with "/" and have more than 1 character. Example: /docs, /hc/en - Exclude Path Filter (optional) Paths (same format as the filter) to be ignored. Example: /internal-portal, /private - Default Tag (optional) Tag applied to the content originating from this source (helps in filtering/searching). 3. Save with Update Source (Save). Important Validations - Invalid URL or with protocol other than http/https causes an error. - Each item in the filter must start with "/"; / alone is not accepted. How does the recurrence (scheduling) work? - You choose Daily / Weekly / Monthly. - The system calculates the schedule considering the local time zone and shows the next estimated run. Examples: - Daily at 06:00 (local) → runs every day at this time. - Weekly (Monday at 07:00) → runs every Monday at the chosen time. - Monthly (on the 15th at 05:00) → runs every 15th of the month at this time. Tip: you can change the cadence at any time; the next execution already considers the new setting. How do I turn on, off, start, or stop synchronization? - Activate/Deactivate: use the toggle in the Status column. - Activated: the source participates in executions (scheduled or manual). - Deactivated: the source is paused, not running manually nor automatically. - Synchronize (manual): click Synchronize. The status changes to Starting/Running. - Stop: while Running, the Stop button appears; click to cancel the current execution. If I deactivate a source, does ClaudIA stop using that content? It depends on the type of deactivation: - 🔄 Disable synchronization: ClaudIA stops updating the content from this source, but still uses what was previously synchronized. - ❌ Remove source: ClaudIA completely stops using that content. If your goal is to control what ClaudIA uses, it's best to remove the source (rather than just disable synchronization) or use source filters on the content screen to isolate and exclude materials from a specific source. How can I see the extracted URLs? Click View URLs. A modal lists all URLs collected for that source. You can open each in a new tab to check the content. What do the statuses mean? Status “badges” appear in the Status column: - Syncing Data collection is in progress. - Synchronized The last execution finished successfully. - Error Something failed. Hover over the badge to see error details. If nothing appears, contact our support team. - Cancelled The execution was interrupted (manually or by system). - Pending No synchronization has been run yet. The page updates the status automatically: faster when there are executions, and at longer intervals when everything is stable. Best practices for path filters - Be specific: use prefixes that represent relevant sections. Example: /hc/en, /help/, /product-x/docs - Avoid generic “/”: this tries to crawl the entire site. - Use “Ignore routes in synchronization” to skip internal or sensitive areas. Example: /admin, /account, /internal-portal Common errors and how to resolve them - “Invalid URL” / “Invalid protocol” Confirm that the URL starts with http:// or https:// and is correct. - “Path must start with ‘/’” Adjust filter items to the format /your-prefix. - Status “Error” Hover over the badge to read the reason (e.g., step failed). Review filters/URL and try to Synchronize again. If the error persists, contact support. - No URLs listed under “View URLs” Check if the path filters include the desired pages. Can I organize or search what has been collected afterward? Yes — the collected contents appear in the Content tab, where you can search by the desired URL and the contents will appear.** If you need to bulk edit something that was incorrectly set, here's how to do it. Can I flag content with different labels/tags within the same source? Yes. You can manually fill in tags (labels) on each synchronized content. When you do this manually, these tags are not overwritten in subsequent synchronizations — only the textual content is updated. This allows organizing a single source into multiple categories or filters. Final tips - Start with one source per area (e.g., Help Center PT/EN separated) and clear filters. - Schedule for times of low traffic on your site. - Review “View URLs” after the first run to ensure only the desired material is being fetched. - Use a Default Tag for easy identification to facilitate searches and reports by source. - How you will see it in Conversation Just like in the Content tab, in the Conversation tab you will also be able to see whether the content is from an external source or not. 📌 List of HTML tags ignored by default During the refinement process, the extraction agent automatically discards various elements that are usually not useful for the knowledge base. These selectors were set to reduce visual and structural noise (menus, ads, comments, forms, etc.), keeping only the main content. Page Structure #footer #header #nav nav footer Scripts and styles sScript style noscript Media svg img audio video Navigation and menus .sidebar .menu .navigation .breadcrumb .breadcrumbs .pagination .pager .page-navigation Ads and banners .advertisement .ads .ad-banner .cookie-banner .cookie-notice .gdpr-notice Social and sharing .social-share .social-buttons .share-buttons Forms and subscriptions .newsletter .subscription .signup-form .search-box .search-form .search-bar Related content .related-posts .recommended .suggestions Comments and discussions .comments .comment-section .discussion Metadata and authorship tags .tag-list .categories .author-bio .author-info .byline .meta .metadata .post-meta Widgets and sidebars .widget .widgets .sidebar-widget Popups and overlays .popup .modal .overlay .lightbox Accessibility and hidden navigation skip-link .screen-reader-text .sr-only .print-only .no-print Accessibility attributes [role='alert'] [role='banner'] [role='navigation'] [role='complementary'] [role='dialog'] [role='alertdialog'] [role="region"][aria-label*="skip" i] [aria-hidden='true'] [aria-modal='true'] Invisible elements .hidden .invisible

Last updated on Jan 07, 2026

Replay: redo and compare a ClaudIA response

https://www.loom.com/share/9ddf19e817ab464ba62f4271ce3c8ff3 Replay re-executes a response ClaudIA already gave, using the current knowledge and configuration, and shows the result side by side with the original response. It's a way to check the effect of a change before trusting it. :::info Replay is only available to users with the Administrator or CX Engineer role, and only on responses generated by ClaudIA's AI. If you don't see the replay button on a ClaudIA response, your user most likely doesn't have that role — talk to an administrator of your account. ::: When to use - After adjusting the knowledge base, to see whether it would change a response ClaudIA already gave. - After changing an instruction or configuration in ClaudIA, to validate the effect in a real conversation. - To understand why ClaudIA answered a certain way — which knowledge base articles it used. Step-by-step Step 1 — Open the conversation Open the conversation with the ClaudIA response you want to re-execute. Step 2 — Trigger the replay Hover over the ClaudIA response and click the replay button (circular arrow icon, "Replay ClaudIA response"). The Replay ClaudIA response panel opens over the conversation. Replay button (circular arrow icon) next to a ClaudIA response Step 3 — Run the replay In the panel, click Run replay. ClaudIA re-executes the response live, from scratch, with the current knowledge. It takes a few moments. Step 4 — Compare the results When it finishes, the panel shows three things: - The verdict at the top: Identical responses or Different responses. - The Context for the responses: which knowledge base articles were used in the original response and in the replay — and whether they changed. - The Original response and the Replay response (now), side by side. Replay comparison panel: verdict at the top, knowledge base context, and original response versus replay response side by side If the response changed, compare the knowledge base articles of the two versions to identify what changed. :::info Replay is only a simulation for comparison: it does not change the conversation or the original response the customer received. ::: Notes - Replay is not available for responses generated by multiple agents (a supervisor that triggers specialist agents). The panel warns you when that's the case. - Responses that didn't come from the AI agent — such as Controlled Flows or manual responses — don't have replay, because there's no agent execution to reproduce.

Last updated on Jun 26, 2026

Improvement Center: how to review and approve ClaudIA's content suggestions

When to use it - When ClaudIA identifies questions from your customers that it doesn't yet know how to answer and suggests new content for your knowledge base. - When the "Centro de Melhorias" (Improvement Center) item appears in the ClaudIA menu with pending suggestions. - Whenever you want to review, edit and enable automatically suggested content — in a single click. What the Improvement Center is The "Centro de Melhorias" (Improvement Center) brings together content suggested automatically from your customers' conversations. When several customers ask something ClaudIA doesn't yet have in its knowledge base — the so-called "no relevant content" conversations —, ClaudIA groups those questions by topic and drafts a suggested answer for you. You review it, edit it if you want and approve it to enable the content in ClaudIA's knowledge base. :::info Suggestions are AI-generated from real conversations — that's why the page itself shows the notice "Generated by AI · review before approving". No suggestion goes live on its own: it is only used by ClaudIA after you approve it. ::: Where to find it In the ClaudIA menu, click "Centro de Melhorias" (Improvement Center) — right below "Conteúdos" (Content). The page opens with: - A summary at the top: "N new suggestions" and the period analyzed ("Generated between … and …"). - A search by question or topic, a sorting option (e.g. Most conversations) and a toggle between the Cards and Table views. - The list of pending suggestions and the "Aprovar todos" (Approve all) button. List of suggestions in the Improvement Center How to read a suggestion Each suggestion includes: - Impact — the "N conversations" chip shows how many conversations fell into that topic without an answer. The higher it is, the more customers that suggestion helps. - Type — New (a topic the knowledge base doesn't cover yet) or Similar (close to something that already exists). - Topic / Tag — the label that groups the content in the knowledge base. - Suggested question and answer — the text ClaudIA proposes. - What customers asked — the real questions that led to the suggestion. Click any of them to open the source conversation. Step by step — review and approve Step 1 — Open the suggestion Click the card (or the pencil icon) to open the "Revisar sugestão" (Review suggestion) box. Review suggestion box Step 2 — Check the source conversations Under "Conversas de origem" (Source conversations), see your customers' real questions. Click any of them to open the full conversation and understand the context behind the suggestion. Step 3 — Review and edit the answer Read the Suggested question and the Answer. Edit whatever you want — both fields are fully editable. If it makes sense, also adjust the Tag, which defines how the content is grouped in the knowledge base. :::warning Some answers come with sections marked as [COMPLETAR] — these are details only you have (a link, a deadline, a path in the system). As long as a [COMPLETAR] remains, the suggestion can't be enabled. Use the "Próximo campo a preencher" (Next field to complete) shortcut to jump straight to the section and fill it in before approving. ::: Step 4 — Decide You have three actions in the review box: - "Salvar e habilitar" (Save and enable) — approves the suggestion and publishes the content to ClaudIA's knowledge base. From then on, ClaudIA starts using it in the answers to your customers. - "Salvar rascunho" (Save draft) — keeps your edits without enabling them (useful when something still needs to be completed). - "Rejeitar" (Reject) — discards the suggestion. It leaves the pending list and doesn't come back. When the answer is complete, the green notice "Ready to enable — this answer will serve your end customers." appears. Step 5 (optional) — Approve in bulk If several suggestions are already complete, use "Aprovar todos" (Approve all) at the top of the list. Suggestions that still have a [COMPLETAR] are left out and stay pending until you complete them. What happens after you approve - The content enters ClaudIA's knowledge base and starts being used in answers — questions on that topic stop falling into "no relevant content". - You'll also find the content under "Conteúdos → Conteúdos Customizados" (Content → Custom Content), with the "Centro de Melhorias" (Improvement Center) badge — and you can filter by the "Centro de Melhorias" source. - Approving isn't final: like any content, you can edit or disable it later under "Conteúdos" (Content). Notes - Suggestions are AI-generated from real conversations and only go live after your approval — always review before enabling. - The period of conversations analyzed appears at the top of the page ("Generated between … and …"). - The impact ("N conversations") helps you prioritize: start with the suggestions that show up most. - Questions about how ClaudIA uses the knowledge base content? See the "Conteúdos | Gerindo o conhecimento da ClaudIA" (Content | Managing ClaudIA's knowledge) category in the Help Center.

Last updated on Jul 22, 2026