Documentation Index

Fetch the complete documentation index at: https://docs.knovvu.com/llms.txt

Use this file to discover all available pages before exploring further.

Version 5.6

  • Published on Aug 19, 2026
  • 5 minute(s) read
  • Suara Atay
 Prev Next

Version 5.6

19 August, 2026

AI Agents

  • Data Table Node - Write Operations:

    • Update Operation:
      The Data Table Node can now update existing rows. Conditions select the rows to update and the values to be written are mapped per column; columns that are not mapped remain unchanged. By default the node returns the number of affected rows, and the "Return affected rows" toggle (off by default) returns the updated rows instead.
    • Upsert Operation:
      A single operation now updates matching rows or inserts a new one when nothing matches. All matching rows are updated, and if there is no match one new row is inserted with the mapped values. The output ports distinguish the two cases: found returns the number of updated rows and notFound returns 1 for an inserted row.
    • Delete Operation:
      Rows matching the given conditions can now be deleted from a Data Table. Deletion is permanent and cannot be undone, and the node returns the number of deleted rows.

  • Data Retention:

    • Automatic Row Cleanup in Data Tables:
      An optional "Automatically delete old rows" setting has been added per Data Table, so tables no longer have to be cleaned up manually before hitting the row limit. Rows that have not been modified for the selected number of days (default 30, range 1–90) are deleted once a day, and the setting is off for all existing tables. Enabling retention or shortening the period asks for confirmation showing how many rows will be removed, and the Data Tables list shows each table's retention period in a new Retention column.
    • Retention for AI Test Evaluation Results:
      AI Test evaluation results are now covered by retention as well, through a separate setting on the Data Retention tab under Tenant Features. The period defaults to 30 days, accepts 1–90 days, and must be shorter than the session retention period. Expired results are deleted by a scheduled job without affecting the session records themselves.

  • Knowledge Base - Website Data Sources:

    • Editing Websites:
      Crawl settings of a website data source can now be edited instead of deleting and re-adding the source. The Main URL stays fixed while all other settings can be changed, and saving asks for confirmation before the website is re-crawled and re-indexed with the new settings. Editing is available on draft versions and is disabled while a website is being indexed.
    • Viewing Indexed URLs:
      A new "Indexed URLs" page shows which pages a website data source actually covers, opened from a magnifier icon on rows whose training is complete. The list is fetched live each time the page is opened, and every URL can be clicked to open in a new tab.

  • MCP Headers Node:
    A new "MCP Headers" node makes it possible to send conversation-specific values to an MCP server, which previously only accepted fixed per-server credentials. The designer selects a server and defines up to 5 header name/value pairs, where values can carry context from the conversation. At runtime, the headers are added only to tool calls of the selected server, alongside that server's own authentication header.

  • AI Compliance Tenant Features:
    A new "AI Compliance" feature group under Tenant Features controls the use of two regulated capabilities, Text Sentiment and Voice Biometrics, which are off by default. While a capability is off, its type is not available in Service Integrations for that tenant. Tenants already using these integrations keep them enabled through a one-time migration during the upgrade.

  • Branding and White-Labeling:

    • Enable Branding Toggle:
      A single "Enable Branding" toggle now controls white-label usage on the Branding Settings page, and it is off by default. While branding is enabled, links to the Sestek documentation are removed from the Knowledge Base retrieval strategy descriptions, and turning the toggle off restores the default logo, favicon and texts.
    • Application Name and Description:
      New "Application Name" and "Application Description" fields let the home page title and its description be customized, and the browser tab title now follows the configured application name.

  • Live Agent Visibility:

    • Agent Names in Conversation Details:
      The name of the live agent who took over a conversation is now shown above their message bubbles on the conversation detail screen, including subsequent transfers between agents, so handovers can be followed per agent. AI Agent names are also displayed above their own bubbles. The information is populated for new conversations only.
    • Agent Details in the Conversational Analytics Integration:
      Live agent id and name are now sent as attached data to Conversational Analytics, so agent-based analysis can be carried out on the CA platform. When a conversation passes through several agents, all of them are included.

  • Agent Builder, Improver and Optimizer:

    • File Attachments:
      Files can now be attached to the Agent Builder, Improver and Optimizer chat panels, so material such as customer flow diagrams can be used as a starting point for a design. PDF, image, JSON, text and Markdown files are supported, up to 5 MB per file and 5 files per request, and uploaded files are not stored.
    • Speech Input:
      Requests can now be dictated instead of typed, using a microphone button in the chat input of the same panels. The button appears when the project has a Speech Recognition integration in use, and the recognition language follows the project's primary language.

Improvements

AI Agents

  • Total Session Duration Widget:
    A "Total Session Duration" card has been added to the Business Metrics tab on the Dashboard page, showing the combined duration of all sessions in the selected date range and channel. This makes the total interaction load measurable independently of the number of sessions.

  • Localized Date Format in Conversations Export:
    Date values on the Conversations screens are now localized according to the interface language, and exported files use the same format.

  • Immediate Activation of Updated License Subscription:
    License subscription settings now take effect as soon as they are saved, so credit consumption continues over the new subscription without republishing the project.

  • Tooltip for the Conversation Hold Message:
    An info tooltip next to the Conversation Hold Message field now states that hold messages are supported for Sync API calls only and cannot be used with the Async API.

  • Published Conversations Only in Dataset Sampling:
    Dataset creation now samples published conversations only, so test conversations made on the draft version no longer skew evaluation results.

  • Progress Indicator in the Assistant Panels:
    The typing indicator in the Agent Builder, Improver and Optimizer panels has been replaced with an animated icon and a rotating progress label, making it clear that a long generation is still running.

  • Minor UI Improvements:
    Several minor improvements have been made to the user interface to
    enhance user experience and usability.

  • Minor Bug Fixes:
    Various minor bugs have been addressed to improve overall system
    stability and performance.