How to Use AI Suggestions

AI suggestions analyze the customer's message and context to generate recommended replies. Instead of typing from scratch, you can review, edit, and send an AI-suggested response.

What AI suggestions analyze:

• Customer message content: What they're asking or reporting
• Conversation history: The last 10 messages in this conversation
• Company context: Your company name and description
• Quick replies: Your existing quick reply templates, so AI can suggest matching ones

Why use AI suggestions:

• Faster responses: Get a draft in seconds instead of typing from scratch
• Consistency: AI maintains the same tone and quality across all agents
• Coverage: AI suggests responses even for unfamiliar questions
• Learning: New agents can learn from AI suggestions while ramping up

AI as a co-pilot, not replacement: AI suggestions are starting points, not final answers. You should always review and edit them before sending. The AI handles the first draft, you add the personal touch.

Note: AI suggestions only appear for the latest inbound (customer) message. If the last message in the conversation is from an agent, no suggestion is shown.

When you're viewing a conversation, AI suggestions appear in the message composer area, typically above or below the text input field.

How suggestions appear: After the customer sends a message, AI generates a suggested response that appears in a bar above or below the message input. This happens automatically — you don't need to trigger it manually.

What you'll see: The suggestion bar shows a preview of the AI-generated response. If AI also identifies a matching quick reply from your templates, it will highlight that quick reply in the quick replies bar.

Tone: The suggestion follows the tone you configured in AI settings (professional, friendly, or casual). Owners can change this in Settings → Replies → AI Settings.

Click to use: Click the "Use suggestion" button to insert the AI response into the message composer. You can then edit it before sending.

Once you find a suggestion you like, click it to insert it into the message composer. Always review and customize before sending.

Review before sending:

• Accuracy: Does the suggestion actually answer the customer's question?
• Tone: Is the tone appropriate for your brand and this customer?
• Completeness: Is any important information missing?
• Personalization: Does it address the customer by name and reference specific details?

Common edits to make:

• Add personalization: "Hi [Customer Name], thanks for your patience..."
• Clarify details: The AI might not have the latest product updates or policy changes
• Adjust tone: Make it more casual for VIP customers or more formal for enterprise accounts
• Add next steps: "I've submitted a ticket for our technical team (reference #1234)..."

Send or add to queue: After editing, either send immediately or save as a draft if you need more time. The suggestion is now your custom response.

Don't see relevant suggestions? If none of the AI suggestions fit, you can type your own response. Also consider giving feedback on why the suggestions missed the mark — this helps improve the AI over time.

AI suggestions require an API key to be configured by your workspace Owner. Without a key, AI features won't be available.

Configuring tone: In Settings → Replies → AI Settings, Owners can adjust the tone of suggestions to match your brand:

• Professional: Formal, business-appropriate language
• Friendly: Warm, approachable tone
• Casual: Relaxed, conversational language

Enabling/disabling: Owners can toggle AI suggestions on or off from the AI settings page. When disabled, no suggestions will appear in conversations.

Common gotchas:

• Suggestions not appearing: Check that AI is enabled in settings and that an API key has been configured by the workspace Owner
• Slow generation: AI suggestions typically take 2-5 seconds. If it's much slower, check your internet connection or AI provider status
• Irrelevant suggestions: Make sure the conversation context is clear. Sometimes vague customer messages produce generic suggestions

Success indicator: When you can respond to customer questions 2-3x faster using AI suggestions — and customers still feel heard and helped — you're getting maximum value from this feature.