Chunks

the chunks feature in pusaka allows you to break large pieces of content into manageable sections, known as “chunks ” this makes information easier to reference, improves retrieval efficiency, and helps maintain clarity when displaying or processing data overview in the chunks section, you can filter chunks by content, title, section, creation type, or state list existing chunks, showing their basic details at a glance create new chunks or edit existing ones to keep the text and metadata accurate deactivate chunks that you no longer want the ai assistant to reference chunks list chunks summary filter content select a content item (e g , “tips02 road race tips”) to show only chunks belonging to that piece of content title search by chunk title section filter by chunk section number(s) creation type filter results by how the chunk was created (e g , by system vs manually ) state specify whether you want to see active or inactive chunks after setting your filters, click apply to refine the list or reset to clear your filters chunk table title brief identifier describing the chunk’s content (e g , “2 perlengkapan yang perlu disiapkan”) state indicates if the chunk is active (used by the ai) or inactive section numerical reference (e g , 1 2 ), useful for ordering chunk creation type shows whether the chunk was generated by system (auto chunking) or manually created by a user chunk detail when you click on a chunk or select new chunk , you can edit or define chunk detail content the parent content item this chunk belongs to (e g , “tips02 road race tips”) title a descriptive name or heading for the chunk section a numeric or labeled indicator to order or categorize the chunk (e g , 1 2 ) state active if the ai assistant will use this chunk as reference during queries; otherwise, inactive if you need to edit manually text token displays the token count for the chunk’s text (e g , 451 ) this is helpful for understanding the size and complexity of the content chunk creation type by system auto generated by the platform’s chunking process manually created or edited directly by a user text the actual content of the chunk keep this concise and focused on a single topic or subtopic to optimize retrieval efficiency deactivate button allows you to switch a chunk’s state from active to inactive, preventing it from being referenced by the ai assistant chunk activation once deactivated, you can edit chunk according to your needs click save and activate if you need this chunk to be referenced again by the ai agent or if needed, you can delete the chunk as well how to manage chunks filter and search use the fields at the top of the chunks page to locate specific items quickly create or edit click new chunk to add a fresh chunk, or select an existing chunk’s row to edit details fill in (or update) the content , title , section , and text fields to keep the information relevant save or deactivate click deactivate in the chunk detail view if you need to temporarily remove a chunk from the ai’s active references and edit chunk according to your needs otherwise, ensure to save your edits so changes persist best practices keep chunks focused each chunk should cover a discrete topic or section, making it easier for the ai assistant to retrieve and present relevant data use descriptive titles clear, concise titles help you and your team manage and find the correct chunk quickly monitor token counts watch the text token metric extremely large chunks could impact performance or clarity regular updates review chunk content periodically to ensure accuracy, especially if source data changes next steps once you’ve configured your chunks check contents to ensure they’re properly categorized and permissioned review reference configuration to confirm these chunks are accessible by the ai assistant for relevant answers consider permissions if certain chunks should only be accessible to specific user roles