Released Enhancements
External Website Content as a Knowledge Source
Admins can now add external websites as content sources that agents can reference when generating responses. This significantly expands what your agent can “know,” especially when critical information is not stored within CivicPlus systems. This enhancement provides more complete, accurate responses and reduces dead ends for users.
Example use cases:
A municipality links to a state website for tax regulations: the agent can now answer detailed tax-related questions using that source
A city references a third-party parks and recreation vendor site: users can get answers about registrations or schedules without leaving the chat
Supplement gaps in your own website content without needing to duplicate or maintain that content internally
Expanded Connected Content Sources
New integrations for Web Evolve and HCMS and improved connect/disconnect controls for content sources have been added. Admins now have more flexibility to curate exactly which content an agent can use. This results in better governance over agent responses and improved accuracy.
Example use cases:
Temporarily disconnect outdated content during a site redesign
Connect multiple systems to create a more holistic knowledge base
Fine-tune which departments or services an agent can speak to
Conversation Search & Filtering Enhancements
We have released an improved ability to search and drill into conversations. This makes it easier to understand what users are asking and how the agent is performing.
Example use cases:
Keyword search for all conversations related to “permits” to identify gaps
Review specific interactions that led to escalations
Citations & Confidence Scores in Responses
CivicPlus Agents now display citations (connected content, external websites, uploaded documents). Confidence scoring has also been added to responses. This provides transparency into where answers come from and how reliable they are.
Example use cases:
A user sees a link to the exact page where the answer came from
Admin identifies low-confidence responses and improves content
Widget Preview & Testing Enhancements
Admins can now preview the chat widget more easily to test agent responses before going live. This reduces risk when launching or updating an agent and provides higher confidence in deployments and fewer production issues.
Example use cases:
Validate how an agent responds to edge-case questions before publishing
Test how newly added content affects responses
Chat Floating Action Button Size
The Floating Action Button (FAB) can now be resized. This creates a smoother, more customizable user experience for end users and admins.
Example use cases:
Adjust widget size for better accessibility
