Knowledge Base
| Topic | Detail |
|---|---|
| Navigation | Admin → Knowledge |
| Purpose | Create and manage AI-powered knowledge bases with Q&A articles surfaced to agents, bots, and self-service portals |
| Technology | Natural Language Understanding (NLU) + Generative AI answer generation |
| Max Knowledge Bases | 500 per organization |
| Max Articles per KB | 15,000 articles |
✅ Verified against Genesys Cloud Resource Center — March 2026
Licensing Requirements
| License | Knowledge Access |
|---|---|
| Genesys Cloud CX 1 | Not included — requires Digital Add-on II or AI Experience tokens |
| Genesys Cloud CX 1 + AI Experience tokens | Access granted without Digital Add-on II |
| Genesys Cloud CX 1 Digital Add-on II | Included |
| Genesys Cloud CX 2 / CX 2 Digital | Included |
| Genesys Cloud CX 3 / CX 3 Digital | Included |
| Genesys Cloud CX 4 | Included |
Required Permissions
| Permission | Purpose |
|---|---|
Knowledge > All |
Full knowledge base administration |
Analytics > Knowledge Aggregate > All |
View knowledge analytics |
Responses > Library > All |
Manage response libraries |
Response Assets > Asset > All |
Manage embedded media assets |
Overview — How Knowledge Base Works
The knowledge base stores question and answer (Q&A) pairs called articles. When a customer or agent asks a question, Genesys Cloud AI uses Natural Language Understanding to find the closest matching article and return the answer.
The knowledge base powers four key touchpoints:
| Touchpoint | Description |
|---|---|
| Agent Copilot | Automatically surfaces relevant articles to agents during live interactions — no manual search required |
| Virtual Agents / Bots | Architect bot flows query the knowledge base and return answers to customers during self-service |
| Knowledge Portal | Customer-facing self-service website — customers search articles, browse by category, or escalate to an agent |
| Messenger | Web Messenger deployments can query knowledge articles in bot conversations |
Step 1 — Create a Knowledge Base
⚠️ Language selection is permanent — it determines which NLU model processes the content. Create separate knowledge bases for each supported language if your contact center is multilingual.
Step 2 — Create Categories & Labels (Optional)
Categories and labels organize articles for easier management and navigation.
Categories
Categories group articles by topic and support nested hierarchies (parent → child).
Labels
Labels are color-coded tags for quick filtering and content reuse across portals.
- Click the Labels tab
- Enter a Label Name
- Select a Label Color
- Click Create Label
💡 March 2026: Knowledge portals can now be configured to return only articles matching specific labels — a single article can be reused across multiple portals with different label filters applied.
Step 3 — Create Articles
Articles are individual Q&A pairs. Each article has one primary question and one answer.
Create a New Article
Import Articles (Bulk)
- Open the knowledge base → Click Import
- Select a .json file containing Q&A pairs
- Genesys Cloud validates the file for errors
- Click Import to confirm — pairs are imported as individual FAQ articles
Step 4 — Format Articles
| Format Option | Description |
|---|---|
| Text Styling | Bold, italic, bullet lists, headings |
| Images | Upload images directly into the answer body |
| Videos | Embed video via URL |
| Rich Text | Full HTML-style formatting for structured answers |
Step 5 — Add Phrasings (Alternative Questions)
Phrasings are alternative ways customers might phrase the same question — used to train the NLU model for better matching accuracy.
- Open an article → click the Phrasings tab
- Add an alternative phrase (e.g.,
How do I change my password?,Forgot my password) - Optional: Enable Autocomplete — generates the phrase as a search prediction when customers type in the portal
- Click Save
💡 More phrasings = better NLU accuracy. Use real customer language, not formal documentation language.
Step 6 — Test Articles
- Open the knowledge base
- In the Test Articles pane, type a test question and press Enter
- The system returns the matched article with a Confidence Percentage
| Confidence Level | Meaning |
|---|---|
| High (80–100%) | Strong match — article surfaces reliably |
| Medium (50–79%) | Acceptable — consider adding more phrasings |
| Low (<50%) | Weak match — improve phrasing or content |
Step 7 — Publish Articles
- Open the article
- Click Publish to publish and continue editing
- Or Publish & Close to publish and exit
⚠️ Unpublished articles are invisible to all touchpoints — agents, bots, and the portal will not see them.
Article Touchpoint Variations
The same article can have different answer content per touchpoint — tailoring responses for agents vs. bots vs. self-service portals.
| Touchpoint | Use Case |
|---|---|
| Agent Assist / Agent Copilot | Detailed internal answer with troubleshooting steps and escalation guidance |
| Bot Flow (Dialog Engine / Digital) | Short, conversational response suitable for bot delivery |
| Knowledge Portal | Formatted customer-facing answer with images and links |
| Messenger | Concise answer for Web Messenger bot conversations |
Third-Party Knowledge Base Integration
Genesys Cloud supports connecting external knowledge systems so their content appears in Agent Copilot, Messenger, and the portal alongside native articles.
| Integration | Sync Type | Notes |
|---|---|---|
| Salesforce Knowledge | Automatic or Manual | Select channels and categories to sync |
| ServiceNow Knowledge | Automatic or Manual | Standard template articles only |
| SharePoint | Via Knowledge Fabric | Configured through Knowledge Configuration |
Adding a Third-Party Source
⚠️ The third-party integration must first be configured in
Admin → Integrationsbefore it appears as a source option here.
Knowledge Configuration (Knowledge Fabric)
Knowledge Configuration defines how knowledge is presented across Genesys touchpoints using a unified layer that can combine multiple knowledge bases and third-party sources.
| Feature | Detail |
|---|---|
| Navigation | Admin → Knowledge → Knowledge Configuration |
| Purpose | Unified knowledge source for Agent Copilot and Virtual Agents — supports generative AI answers |
| AI Answer Generation | Combines content from multiple relevant articles into a single dynamic response |
| Virtual Agent Use | From February 2026 — bots use either Knowledge Workbench V2 or Knowledge Fabric — one at a time, selected in Architect |
💡 February 2026: Bot authors can now select a Knowledge Fabric configuration in Architect and control generative answer mode and bias settings per flow. Existing rules-based bots using Workbench V2 are unaffected.
Agent Copilot & Knowledge
Agent Copilot uses the knowledge base to automatically surface articles to agents during live interactions.
| Feature | Description |
|---|---|
| Automatic Surfacing | Articles appear in the Agent Copilot panel based on conversation context |
| Answer Highlighting | Highlights the most relevant passage within a lengthy article |
| Knowledge Sharing | Agents can send articles directly to customers with one click |
| Interaction Summaries | AI-generated summaries at end of interactions using conversation + knowledge context |
| Wrap-Up Code Suggestions | Copilot suggests wrap-up codes based on the conversation |
Knowledge Portal
The Knowledge Portal is a customer-facing self-service website powered by the knowledge base.
| Feature | Description |
|---|---|
| Article Search | Customers search using natural language — NLU finds the best match |
| Category Browsing | Customers navigate by category hierarchy |
| Chatbot Access | Customers can escalate to a bot or live agent directly from the portal |
| Label Filtering | Portals can filter articles by label (March 2026 feature) |
| Customization | Configurable colors, messaging, and branding |
Knowledge Analytics
| Metric | Description |
|---|---|
| Top Articles | Most frequently accessed articles by agents and customers |
| Search Success Rate | Percentage of searches returning a useful result |
| Search Failure Rate | Searches returning no relevant result — indicates content gaps |
| Confidence Scores | AI match confidence per article returned |
| Containment Rate | Issues resolved through self-service without agent escalation |
Interview Cheat Sheet
| Question | Answer |
|---|---|
| Max knowledge bases per org? | 500 |
| Max articles per knowledge base? | 15,000 |
| What must happen before an article is live? | It must be Published |
| What is a Phrasing? | An alternative way a customer might ask the same question — trains the NLU model |
| What does Confidence Percentage mean? | How closely a query matches the article — higher = better match |
| What are the four main touchpoints for knowledge? | Agent Copilot, Virtual Agents/Bots, Knowledge Portal, Messenger |
| What is the Knowledge Portal? | A customer-facing self-service website for searching and browsing articles |
| What third-party sources can be integrated? | Salesforce Knowledge, ServiceNow Knowledge, SharePoint (via Knowledge Fabric) |
| What is Knowledge Fabric / Knowledge Configuration? | A unified knowledge layer combining multiple sources with generative AI answers for Agent Copilot and Virtual Agents |
| Can a bot use both Workbench V2 and Knowledge Fabric? | No — one at a time, selected in Architect per bot flow (February 2026) |