Release Date: Jul 16th 2026
Build: 8.0.0.20895
Install ID: sw_800

The 8.0.0 release dramatically enhances the AI capabilities of the platform with some exciting new features, in particular adding AI Agents that can answer policy questions and provide assistance while using the system. We've also fixed a number of bugs and made other miscellaneous enhancements. 

Cloud hosted systems will be updated automatically in the coming days. For on-premise, learn how to upgrade here.

Note: This replaces the 20880 (July 14th) build given to select clients. Please install sw_800 again to resolve an issue with printing images in PDFs.

Features:

1. Introducing AI Agents! Currently mostly available to admin users (with a special Policy Support Agent for end-users), agents are built-in chatbots that specialize in different areas of Staff.Wiki. There are Built-In Agents for policy management, learning management, incident management, compliance and risk, configuration, asset management and more. What's more - you can create your own agents, too, that specialize in different areas of your organization. You can add as many agents, or sub-agents, as you need.
    
   1. A Policy Support Agent is available to end users that can answer end user questions about any policy in your wiki. It'll appear in the familiar Questions box, or also when they enter a question into the search box on the wiki page (more info below).
   2. Admin users will find multiple agents in the "Agents" tab (the Agent Dashboard), invisible for end users. You'll also be able to access these same agents when using other areas of the application, as a chat box in the bottom right hand corner. For example, when on an incident, you'll see an "Incident Reports Expert" box in the lower right corner where you can discuss the incident report you are viewing.
   3. Agents have a list of skills that you can manage, telling it how to perform different actions or go about answering specific questions.  See Agent Training and Skills.
   4. Each agent has access to its own hidden Agent Browser and has been trained specifically on using Staff.Wiki, but can also be trained on other sites. Agents can easily be configured to have no access to a browser, access only to Staff.Wiki, access to white-listed sites, or access to any site - for maximum security.
   5. Agents can actually take actions on your behalf, whether updating a policy or collecting information for an incident investigation.
   6. Agents are self-improving (see Agent Training and Skills). If they are asked (or asked to do) something they are unsure of, or unable to complete, they will open a side-chat with their supervisor to ask for guidance - notified via email. Once the answer is given, the agent will automatically update its skills so it knows how to handle that question or task the next time.
   7. Agents that are marked as "in training" will always ask their supervisor to check (or correct) any forms/updates they fill out before submitting.
   8. If an agent is asked to do something outside of their expertise, they will confer with another agent. This agent collaboration allows for optimal use of the available context-window size which can save costs.
   9. Agents can create images and infographics to help provide explanations or guidance.
   10. Agents can produce documents and spreadsheets that can be downloaded.
   11. You can upload documents and spreadsheets and reference them in your conversations.
   12. Agents can be associated with a user login that is specific to them. Any assignments made to that user will be processed automatically by the associated agent in the background. 
   13. Because of security concerns, agents that can perform actions or access the browser can only be accessed by an admin user. Agents can be set to record the screen clicks performed by an agent for auditing purposes.
   14. Cost saving features let you control the level of reasoning and context-window usage allowed on each agent, along with setting maximum conversations per user per day, and maximum number of turns in a conversation.
   15. Note: Use of agents requires that Enhanced AI is enabled, which requires signing up for an OpenAI Platform account. AI usage will incur token usage fees as per OpenAI's current pricing.

2. The Questions box (on policies) and the Search Box in the wiki can now be configured to use a special, more restricted, AI Policy Support Agent to answer questions about either the policy being viewed, or any other policy the user has access to in the system. To control AI service usage (and thus cost), you can restrict how many chats a user can have a day, how long those chats are, what level of AI reasoning is used, and how large the context would be when answering those questions.

3. The Quiz Results report on the User tab now also lists all failed and untaken questions.

4. Moving articles no longer changes the "Last Modified" date that shows in the statistics for an article. This is now reserved only for actual updates to content.

5. When a user group is set as an approver, but the editor is a member of that group such that the approval should be auto-accepted, the user group members are no longer notified of a pending review.

6. You can now report incidents by email. Once configured, the email address of the email sender is used to look up the department to send to. The user must be in a department, and the department must have a default incident assignee set to receive the incident for further processing. Once posted, a link to the incident is sent back to the user for further tracking. Only the subject and the body of the email are used in the incident report, other details can be provided after it is recorded in the system.

7. Some reports in the system had a hard limit of 500 records in the output. This limit has been removed.

8. You can now force-update change metadata as an admin, which can be useful for importing histories of policies.

9. You can now set a dark mode custom logo. Previously only the light mode logo could be updated, and the dark mode would show as black and white. Go to the Configuration tab, under Main Settings, you can upload the file (PNG or SVG) to the Override Logo (Dark Mode) field.

10. The side navigator (contents link) in mobile view now has a larger font so it's easier to navigate.

Fixes:

1. Fixed a low impact/benign XSS security issue (related to Explanations when publishing).
2. Fixed an error that is displayed when a user is inserted with a department and user group set that have the same name, but the department is set to create user groups.
3. Fixed an issue with pasting lists that contain margins after the bullet, which caused the editor to break with cursor positioning.
4. Fixed another low impact/benign XSS security issue (related to language switching).
5. Fixed an issue with repeating tasks where, in some conditions (especially on the legacy cloud edition) you would not receive assignments for subsequently repeating tasks.
6. Fixed an issue with the metadata encoding on articles, where sometimes certain characters would be encoded incorrectly.
7. Miscellaneous cosmetic fixes.
8. In the history view, changes/diffs from one version to another would sometimes show incorrect section block changes because of the use of indentations, which made it look as if a whole section was removed. The diffing engine now ignores those indentation changes.
9. Fixed an issue with background downloading of sample policy library updates. Occasionally this process would sometimes skip certain updates to sample articles because of a timing issue that has been resolved.