Before you can use the Agento Intelligence system for Govintra, you must have a valid API key from Agento Digital, and you must have completed the security integration process. Access to the AI system is blocked without prior security authentication and API key.
Agento Intelligence is an add-on to Govintra and requires subscription.
Getting Started
Agento Intelligence adds an AI-powered assistant to your intranet. Staff can ask questions about intranet content and receive answers drawn from your published tasks, pages and documents. The plugin settings are found under Options → Agento Intelligence in the WordPress admin.
Authentication
Before configuring any other settings, you must connect to the Agento Intelligence backend service.
API Key
Your unique API key provided by Agento Digital. This authenticates your site with the AI backend. Keep this confidential — do not share it.
Host URL
The base URL of your Agento Intelligence backend service. This is provided during setup and should not normally need changing.
Access Mode
Controls who can use the AI assistant:
- Logged in users only — Only authenticated WordPress users can ask questions. This is the recommended setting for most intranets.
- Public access, not logged in — Allows unauthenticated access using a public token. Useful for demo sites or public-facing help pages. Includes per-IP rate limiting.
- Private access, not logged in — No authentication required. Only use this if your intranet is already protected at the network level (e.g. behind a VPN).
Public Token
Only required when using “Public access” mode. Click Generate token to create a secure token automatically. This token must be included in the page URL or request header for unauthenticated users.
Settings
Content Indexing
These options control which intranet content is searchable by the AI assistant.
Index Tasks
When enabled, all published tasks and guides are included in the AI’s knowledge base. This is the primary content type for most intranets and is enabled by default. Content is automatically re-indexed whenever a task is saved.
Index Pages
When enabled, specific pages can be included in the knowledge base. Unlike tasks, you must explicitly select which pages to include.
Pages to Include
Select the top-level pages you want indexed. All child pages beneath a selected parent are automatically included. For example, selecting your “About us” page will also index all pages nested beneath it.
Index Documents
When enabled, documents in the Media Library (PDFs and Office files) are indexed by their title, description and filename. Note: the actual file contents are not extracted — only the metadata WordPress stores. This helps the AI recommend relevant documents when answering questions.
Behaviour
Collect Feedback
Adds thumbs up/thumbs down buttons beneath AI responses. Users can optionally leave a text comment. Feedback is stored in the query log for review.
Feedback Style
Choose between a branded feedback interface or a minimal plain style.
Track Analytics
When enabled, questions are tracked in your analytics platform (Google Analytics or Matomo). This helps you understand what staff are searching for.
Log Questions
Stores all questions and AI responses in a database log, viewable under Tools → Agento Intel Logs. Logs are automatically purged after 30 days. This is useful for identifying gaps in your content — if staff repeatedly ask questions the AI cannot answer, you may need to create or improve a task page.
Allow History
Shows the user’s recent questions in the chat interface, allowing them to quickly revisit previous queries.
Batch Size
Controls how many content items are sent to the backend per indexing request. The default of 10 works well for most sites.
Cache Period
Set in hours. When a user asks the same question that was asked recently, the cached response is returned instantly without calling the AI backend. Set to 0 to disable caching. A value of 1–4 hours is recommended to balance freshness with performance.
Personalisation
These options let you brand the AI assistant to match your intranet’s identity.
Thinking Text
The message displayed while the AI is processing a question. Default: “Checking intranet content…”. You might change this to something like “Searching ACMEnet…” or “Looking that up for you…”.
Query Examples
Provide example questions that animate as placeholder text in the search input, giving users inspiration for what to ask. Separate multiple examples with a pipe character (|).
Example: How do I book a meeting room?|What are the opening hours?|Who do I contact about IT issues?
Disclaimer Text
An optional message shown at the bottom of every AI response. Use this to remind users that AI responses are generated and should be verified. For example: “This answer was generated by AI. Always check the linked source page for the most up-to-date information.”
Brand Colours
Two colours used for the AI assistant interface elements — gradients, buttons and accents. These default to cyan and purple but should be adjusted to complement your intranet’s brand colours.
Personalise AI Chat
When enabled, chat messages display name badges — the logged-in user’s name and a custom AI assistant name.
AI Assistant Name
The display name for the AI bot in the chat interface. For example: “ACMEnet Assistant”, “Ask HR”, or “Bertie”.
System Prompt
Advanced setting. The system prompt instructs the AI on its role, tone and behaviour. The default works well for most sites. Only change this if you need the AI to adopt a specific personality or follow particular guidelines. For example, you might instruct it to always respond in formal language, or to ask for feedback after responses.
Writing Tools
AI Rewriter
Adds an AI-powered rewriting tool to the WordPress content editor toolbar. Authors can select text and apply style transformations — useful for ensuring consistent tone across the intranet.
Rules
Define custom rewriting rules that appear in the editor toolbar dropdown. Each rule has a title (shown to users) and a prompt (instructions sent to the AI). Common examples:
- Simplify — “Rewrite this text in plain English, avoiding jargon”
- Make Professional — “Rewrite in a formal, professional tone”
- Summarise — “Rewrite as a 50 word summary”
- Translate — “Translate this text into French”
System Status
The system status tab provides maintenance controls for the content index.
Reindex All
Rebuilds the entire vector index from scratch. This re-processes all selected content types (tasks, pages, documents) and sends them to the backend. A progress bar shows completion status. Use this after making significant changes to your indexing settings, or to initialise your index.
Cancel
Stops a reindex operation that is currently in progress.
Purge Index
Permanently deletes all indexed vectors from the backend. Use this with caution — after purging, the AI assistant will have no knowledge of your content until you run a fresh reindex.
Shortcodes
Three shortcodes are available for embedding AI features into pages:
The main AI chat interface. Place this on a dedicated “Ask a question” page. Optional attributes:
show_filters="no"— Hide the content type and category filterstype="task"— Pre-filter results to a specific content typecategory="hr"— Pre-filter to a taxonomy categorytag="policy"— Pre-filter by tag
A simpler vector search interface without AI-generated answers. Returns a list of relevant pages ranked by similarity. Faster than the full RAG chat. Accepts the same filter attributes.
You must be logged in to use this tool.
You must be logged in to use this tool.
An optical character recognition tool allowing users to upload images and extract text. Only available to logged-in users.
WP-CLI Commands
For sites with SSH access, the following commands are available:
wp agento-intel reindex— Full reindex (supports--types,--dry-run)wp agento-intel purge— Purge all vectorswp agento-intel status— Show configuration summarywp agento-intel stats— Backend index statisticswp agento-intel ping— Check backend connectivity