Serve customers faster and smarter with an AI-powered live chat assistant that gives your agents accurate answers and clear next-step guidance - all based on your own brand knowledge.
At Talkative, we call our agent copilot Navi. So if you see the name Navi within your console or documentation, don’t worry, it’s simply the internal name for your AI copilot.
To use Copilot effectively, you’ll need to build your Knowledge Base first. Copilot relies entirely on this content to provide accurate answers and guidance, so make sure your brand information are added before enabling it.
⚙️ Setting up your Copilot Config
Step 1: Access Copilot Configs
- Via Settings in the left side menu, click ‘Copilot Configs’ under ‘Integrations’
- Hit ‘Create Copilot Config’
Step 2: Complete Copilot Form
- Add a name / label.
- Choose the Queues that Navi will be working for.
- Select the Knowledge Base that Navi will use.
- Pick your preferred model of AI.
Now you can enable the features of Navi…
- AI Message Auto Complete is like a predictive text feature based on what the agent has started to type.
- Navi, the AI Assistant, when enabled, will present an internal chatbot for the agents to use.
- AI Message Suggestions, when enabled, will send reply suggestions to the agent based on the customers interactions.
- You can make these settings a default for the whole company.
- Hit Create Copilot Config.
Step 3: Enabling AI Reword
Separate from your Copilot configuration, you’ll also want to visit the AI Reword Message setting in the Agent Experience menu.
Enabling this feature gives agents access to quick rewording options, such as spelling & grammar fixes, shorter or longer phrasing, and tone adjustments like “friendlier” or “more professional” directly from the chat input bar.
✅ Copilot Functionality
AI message auto complete
When enabled, AI suggests messages to the agent based on what they have already written.
Individual users can disable this feature by going to User Profile > User Configs > Disable
Navi, the AI Assistant
When enabled, this internal-facing chatbot will be available to agents to answer questions based on the interaction and the knowledge base. Navi also leverages more general knowledge available to the model.
AI message suggestions
When enabled, the agent will receive suggestions based on the question the customer has asked. When you accept interactions in the console, you will be given a message suggestion based on your knowledgebase:
The suggestion will take into account the entire transcript/conversation.
If no appropriate response is returned from the model, the agent will not receive a suggestion.
Canned messages will appear below the AI answer once you start typing:
Further Examples:
In the Copilot panel, Navi automatically suggests an answer pulled directly from your Knowledge Base.
The agent has two options:
- Fill Chatbox – This inserts the suggested response into the chat input area, where the agent can review and edit it before sending.
- Send Message – This sends Navi’s suggested answer immediately, without making any changes.
Below this, the Navi panel in the bottom-right corner provides additional AI-powered tools. Here, the agent can:
- Summarise the current interaction
- Generate a Next Action recommendation
- Or ask Navi a direct question using the text field at the bottom (e.g., to clarify details, get guidance, or request additional information)
This section helps agents combine fast AI-powered responses with the flexibility to review, edit, or query Navi as needed.
In this example, Navi is supporting the agent in three ways:
- Suggested answer (top right): Navi pulls a relevant response from your AI Knowledge Base, ready to insert or send.
- Next actions (bottom right): Navi recommends helpful next steps based on the ongoing conversation.
- Message autocomplete: As the agent begins typing, Navi offers a completion option that aligns with best-practice phrasing.
Navi combines your Knowledge Base with the live interaction transcript to generate accurate, context-aware guidance for the agent.
Data: Only the message transcript is sent to OpenAI, pre-chat form data and other interaction metadata is not used. OpenAI does not store any messages or responses. Please familiarise yourself with OpenAI's privacy policies before using the system. https://openai.com/policies/api-data-usage-policies