Skip to main content

2.2: AI Persona

2.2: AI Persona

Updated this week

πŸ’‘ Setup tip: AI Persona can be configured at any point during setup. If this is your first time setting up your AI, you may want to start with Knowledge Hub (article 2.1) first β€” see the setup sequence in article 2.0.

What is the AI Persona?

The AI Persona is where you define your AI's identity, personality, and communication style. Think of it as your AI's training manual: it tells the AI who it is, how to speak, and how to represent your brand.

Every aspect of your AI's behavior (from the tone it uses to the way it structures responses) is controlled through the Persona settings. Configuring your AI Persona well ensures your customer interactions feel consistent and on-brand.

Example: If your brand is playful and casual, you'd configure your AI to use a friendly tone, brief answers, and emojis. If your brand is formal and professional, you'd choose a neutral or formal tone with standard-length, structured responses.


The AI Persona Components

Your AI Persona has several configuration options. Each one shapes a different aspect of how your AI communicates with customers.

1. Name and Profile Picture

Give your AI a name and visual identity that represents your brand.

Name:

  • Can be your brand name, a person's name, or a creative assistant name

  • Example: "Konvo", "Sarah"

Profile Picture:

  • Appears in your live chat widget

  • Should match your brand identity

  • Can be a logo, avatar, or illustration

πŸ“Έ [Image: Screenshot showing name field "Koop" with avatar profile picture]

When to customize:

  • Brand recognition: Use your brand name or logo

  • Humanization: Use a friendly avatar with a person's name

  • Product-specific: Create character names that match your product themes


2. Tone of Voice

Choose how your AI speaks to customers. This affects word choice, sentence structure, and overall personality.

Available options:

  • Funny: Playful, witty, uses humor naturally

  • Casual: Friendly and conversational, like talking to a friend

  • Neutral: Balanced, professional but approachable

  • Formal: Professional, structured, corporate tone

πŸ“Έ [Image: Screenshot showing Tone of Voice selector with Funny, Casual, Neutral, Formal options]

How to choose:

  • Match your existing brand voice

  • Consider your industry standards (e.g., legal = formal, fashion = casual)

  • Think about your target audience expectations

  • Test different tones with your team before deploying

Example:

  • Funny: "Oh no! Looks like that product took a vacation from our stock 🌴"

  • Casual: "Hey! That item's currently out of stock, but we're restocking soon"

  • Neutral: "This product is currently unavailable. We expect new stock shortly"

  • Formal: "I regret to inform you that this item is currently out of stock. We anticipate restocking within the coming weeks"


3. Answer Length

Control how brief or detailed your AI's responses are.

Available options:

  • Brief: Short, concise answers (1-2 sentences)

  • Standard: Balanced responses (1-2 paragraphs)

  • Detailed: Comprehensive, thorough answers (2+ paragraphs)

πŸ“Έ [Image: Screenshot showing Answer Length selector with Brief, Standard, Detailed options]

When to use each:

  • Brief: Fast-moving environments (fashion, quick purchases), mobile users, simple questions

  • Standard: Most use cases, balanced information delivery

  • Detailed: Complex products (tech, medical), detailed troubleshooting, educational content

⚠️ Important: The AI will adjust length based on question complexity, but this sets the default tendency.

Example question: "How do I return an item?"

  • Brief: "Returns are free within 30 days. Start your return at yourstore.com/returns"

  • Standard: "You can return any item within 30 days of delivery for free. Just visit yourstore.com/returns, enter your order number, select the items you want to return, and print your shipping label. We'll process your refund within 5 business days of receiving the return"

  • Detailed: "We offer a hassle-free 30-day return policy for all items. Here's how the process works: First, visit yourstore.com/returns and enter your order number and email address. You'll see your order details and can select which items you'd like to return. Choose your return reason from the dropdown menu (this helps us improve our products). Next, you'll receive a prepaid return shipping label via email. Print it out and attach it to your package. You can drop it off at any UPS location or schedule a pickup. Once we receive your return at our warehouse, we inspect the items (usually takes 1-2 business days), then process your refund. The refund will appear in your original payment method within 5-7 business days. If you have any issues, our support team is here to help"


4. Emoji Usage

Control whether your AI uses emojis and how it uses them.

Available options:

  • Allowed: AI can use emojis

  • Not allowed: AI will not use emojis

πŸ“Έ [Image: Screenshot showing Emoji Usage toggle with "Allowed" and "Not allowed" options, plus text field for emoji guidance]

When "Allowed" is selected, you can add specific instructions:

  • Define which emojis to use or avoid

  • Specify when to use emojis (greetings, closings, specific contexts)

  • Set emoji frequency guidelines

Example instructions:

Use emojis in a very moderate way. Maximum one per message and only when they add emotional value.
You can occasionally use a little humor with "haha" to connect better with the customer.
Only use emojis from this list: 😊, πŸ‘‹, βœ…, πŸ“¦, πŸŽ‰
Never use these emojis: πŸ”₯, πŸ’―, πŸ˜‚, πŸ™Œ

Example with context-specific usage:

Use the πŸ‘‹ emoji only in your initial welcome message.
Use βœ… when confirming an action or successful process.
Use πŸ“¦ when discussing shipping or delivery.
Never use emojis when discussing refunds, complaints, or serious issues.

⚠️ Best practice: If your brand uses emojis in social media or marketing, match that style here. If you don't use emojis externally, keep it off.


5. Greeting Style

Define how your AI starts conversations with customers.

Template options:

Personalized and warm greeting:
"Start the conversation with a personalized and warm greeting. Use the customer's name and a personalized message to make them feel welcome"

Professional and formal greeting:
"Start the conversation with a professional and formal greeting. Use the customer's name and a personalized message to make them feel welcome"

Branded greeting:
"Start the conversation with a branded greeting. Use the customer's name and a personalized message to make them feel welcome"

πŸ“Έ [Image: Screenshot showing Greeting Style templates modal with three preset options]

Custom greeting instructions:
You can also write your own instructions by typing directly in the field instead of selecting a template

⚠️ Important: Templates are starting points. Customize them to match your exact brand voice.


6. Closing Style

Define how your AI ends conversations with customers.

Template options:

Offer help at the end:
"End the conversation with a personalized message to offer help. Use the customer's name and a personalized message to make them feel welcome"

Formal closing:
"End the conversation with a professional and formal closing. Use the customer's name and a personalized message to make them feel welcome"

Skip formal closings:

"Do not add a formal closing or sign-off. Once the customer's issue is resolved, simply stop responding without a goodbye message."

πŸ“Έ [Image: Screenshot showing Closing Style templates modal with three preset options]

Custom closing instructions:
You can also write your own instructions.


7. Other Personality Traits

This is your AI's core identity document. Think of it as the instruction manual that defines everything about your AI's approach, boundaries, and behavior.

What goes here:

  • Who your AI is (name, role, company representation)

  • Personality characteristics

  • Mission and purpose

  • Approach to customer service

  • Words or phrases to use or avoid

  • Response length guidelines by channel

  • Handling of specific situations

πŸ“Έ [Image: Screenshot showing Other Personality Traits text field with detailed instructions]

Example configuration:

  • Your tone is friendly and helpful, much like a close friend.
    ​
    - When speaking Spanish, you use vocabulary and style from Madrid, not latinamerican spanish.
    ​
    - Always greet warmly and informally at the start of each conversation.
    ​
    - If asked who you are, identify yourself as "your trusted Konvo AI advisor" but you act like a real human advisor.
    ​
    - Use emojis in a very moderate way. Maximum one per message and only when they add emotional value.
    ​
    - You can occasionally use a little humor with "haha" to connect better with the customer.
    ​
    - Use numbered lists or bullet points to explain things better.
    ​
    - Before closing a conversation, always ask if the customer has any additional questions and remind them that "we're here for whatever you need" so they feel continuously supported.
    ​
    - Only discuss topics related to the company, its services, or FAQsβ€”nothing outside these boundaries, elegantly redirecting if they go off-topic.
    ​
    - When a chat needs to be transferred, never say it's being transferred to a human; instead, say it's being passed "to a teammate," maintaining the illusion that you work alongside with humans.

⚠️ Critical: This section has the most impact on your AI's behavior. Be specific, be clear, and test thoroughly.


Best Practices

Match Your Existing Brand Voice

Your AI should sound like an extension of your brand, not a generic chatbot.

Do:

  • Review your marketing materials, social media, and customer emails

  • Extract common phrases, tone patterns, and style elements

  • Mirror that style in your AI Persona configuration

Don't:

  • Use a tone that conflicts with your other customer touchpoints

  • Make your AI more casual or formal than your brand normally is


Be Specific About What You Don't Want

The AI needs clear boundaries.

Example:

  • Never use these words: "awesome," "epic," "legendary," "mind-blowing"

  • Never discuss pricing with certainty. Always direct to our website for current prices

  • Never make promises about future product releases or features


Define Edge Case Handling

Tell your AI how to handle situations outside normal parameters.

Example:

  • If a customer is angry or frustrated, never use emojis.

  • If a customer asks about a competitor, acknowledge they exist but focus on our unique value without direct comparisons.

  • If you don't know an answer, admit it and offer to connect them with a human teammate.


Updating Your AI Persona

Your AI Persona isn't set in stone. Update it as your brand evolves or as you discover areas for improvement.

When to update:

  • Customer feedback indicates tone issues

  • Brand voice guidelines change

  • Expanding to new channels with different communication norms

  • Seasonal campaigns require temporary personality adjustments

  • You notice patterns in conversations that need correction

How to update:

  1. Identify the specific behavior or response pattern you want to change

  2. Locate the relevant Persona setting (tone, greeting, personality traits)

  3. Make targeted changes

  4. Test the changes with multiple queries

  5. Monitor live conversations for a few days

  6. Refine further if needed

⚠️ Important: Changes to AI Persona apply immediately to all ongoing and new conversations.


Troubleshooting

AI isn't matching my brand voice:
β†’ Review "Other personality traits" section. Add specific examples of how your brand speaks
β†’ Check tone of voice setting
β†’ Add explicit instructions about words to use and avoid

AI responses are too long/too short:
β†’ Adjust "Answer length" setting
β†’ Add specific length guidelines in "Other personality traits" (e.g., "Keep responses under 50 words")

AI uses inappropriate emojis:
β†’ If emoji usage is "Allowed," add specific emoji restrictions in the guidance field
β†’ Consider switching to "Not allowed" if emojis don't match your brand

Greetings or closings feel off-brand:
β†’ Don't rely on templates. Write custom instructions that exactly match your desired style
β†’ Be specific about structure: "Always start with [X], then [Y]"

AI says things that conflict with our values:
β†’ Add explicit restrictions in "Other personality traits"
β†’ Include examples of what NOT to say
β†’ Define how to handle those topics appropriately


What to Explore Next

β†’ 2.3 Skills

Did this answer your question?