Skip to main content

Documentation Index

Fetch the complete documentation index at: https://setup.cevro.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Purpose

At Cevro, our mission is to deliver the highest quality player support. Too often, outsourced support agents or overwhelmed internal teams don’t have the time to be as patient and empathetic as players deserve. Not so with Cevro. Our default agent is designed to be friendly, empathetic, and helpful. However, we also empower brands to customize their agents as if they were hiring a team of human agents. The goal is to create agents that sound neither impersonal nor robotic. Instead, your AI agent should feel like a warm, humble, polite, and empathetic team member—friendly, informative, helpful, and positive.

​
Accessing Agent Configuration

Configure your AI agents at app.cevro.ai/agents. Here you can create new agents, customize their personality, tone, and style, and manage your suite of agents handling player conversations.

​
Agent Personality & Tone Calibration

Just like hiring human support agents, you want to ensure your AI agents reflect your brand’s unique voice and values. Cevro provides multiple layers of customization to help you calibrate agent personality, communication style, and behavior.

​
Agent Identity & Basic Configuration

Before configuring personality and tone, set up your agent’s basic identity:
Give your agent a recognizable face. You have two options:AI-Generated Photos: Click “Generate Photo” to create an AI-generated avatar based on your agent’s background and personality.Custom Upload: Click “or upload” to use your own image. Supports JPEG, PNG, WebP, and GIF formats. Great for brand-approved headshots or consistent styling across your team.
When an avatar already exists, hover over it to see options: view full-size, upload a replacement, or regenerate with AI.
Important - Help Desk Integration Note: For most help desk integrations (Intercom, LiveChat, Zoho SalesIQ, Respond.io), you’ll need to manually download the avatar and upload it to your help desk. Otherwise, players will see whatever avatar is assigned to the agent seat in your help desk, not the Cevro one.Exception: Zendesk allows Cevro to control the avatar directly.
The main name your agent will use when introducing themselves and signing messages.Important for Help Desk Integrations:
  • If your agent is connected to LiveChat, Zoho SalesIQ, or Intercom, ensure the primary name matches the agent name in your help desk
  • For LiveChat, we sync this automatically
  • For other channels, you may need to match it manually to ensure proper attribution
Add up to 10 alternative display names. On supported channels, Cevro randomly picks one name from this pool (including the primary name) at the start of each conversation — creating the appearance of a larger support team while using a single AI agent with the same persona and training.Supported channels: Zendesk, Respond.io, CevroWidget (Web)
Not available on LiveChat, Zoho SalesIQ, or Intercom — these channels require the agent name to match the help desk agent seat exactly.
Configure a custom greeting that your agent sends when starting a new conversation.Pro Tip - Message Variations: You can provide multiple variations of your welcome message in the same field, and Cevro will randomly select one for variety:
Hey there! 👋 I'm Sophie. How can I help you today?
---
Hi! I'm Sophie from support. What can I help you with?
---
Hello! Sophie here. I'm ready to assist you!
This keeps your greetings fresh and prevents repetitive messaging.
Configure how your agent refers to themselves in different languages or contexts.Options:
  • Male - Agent uses masculine pronouns
  • Female - Agent uses feminine pronouns
  • Auto - Cevro infers pronouns from the agent’s background/name
If you’ve written a detailed background, “Auto” works well as Cevro will intelligently detect the appropriate pronouns.
For LiveChat integrations only: Link your Cevro AI agent to a specific LiveChat agent seat.This ensures:
  • Proper conversation attribution
  • Seamless handoffs between AI and human agents
  • Accurate reporting in both Cevro and LiveChat dashboards
The AI agent will appear as the linked LiveChat agent in your LiveChat interface.Connection Status Indicator: Once linked, you’ll see a real-time health indicator showing whether the agent’s connection is working:
  • Green dot = Connected and ready to handle chats
  • Red dot = Connection issue (with specific error message)
If you see a red indicator, the most common fix is to re-authenticate the agent by clicking “Disconnect” and linking again. Use the “Retry check” button to refresh the status after making changes.
Cevro supports two types of agents, each serving a distinct purpose:AI Agents (default)
  • Fully autonomous AI-powered support agents
  • Configurable personality, tone, and behavior
  • Handle player conversations automatically using AIPs
  • Automatically assigned to new tickets
  • This is what you’re configuring when you create an agent on the Agents page
Human Agents (QA/Scorecard feature)
  • Available only for Intercom help desk integrations
  • Used specifically for the Scorecard QA feature — they do NOT handle automated conversations
  • When a human operator closes a conversation in Intercom, Cevro syncs it and scores it using your configured scorecard
  • Created through Settings → Scorecards, not through the standard agent creation flow
Human agents are not a replacement for AI agents. They exist solely to enable QA scoring of your human support team’s conversations. They will never be assigned to handle automated conversations — only AI agents do that.
Human agents are part of Cevro’s advanced QA capabilities. See the Scorecards section in Settings for more details on scoring human agent performance.
Track your agent’s performance with key metrics displayed on the agent profile:CSAT Score - Customer Satisfaction rating based on player feedbackTicket Count - Total number of conversations handled by this agentOperator ID - If linked to a help desk (LiveChat, Intercom, etc.), shows the corresponding agent ID in that systemThese metrics help you understand agent performance and identify areas for improvement.

​
Agent Background & Backstory

Give your agent a personality and history that players can connect with. The background field allows you to create a rich backstory for your agent that can be referenced in conversations when players ask about them. What to include in an agent’s background:
  • Personal details (age, location, interests)
  • Professional background and expertise
  • Personality quirks and preferences
  • Why they work in player support
  • Fun facts that make them relatable
Example Background: Sunny is a 31-year-old female hospitality expert turned casino enthusiast. Originally from Miami, she spent years as a resort concierge, mastering the art of making every guest feel special. Her love for vibrant energy led her into the online gaming world, where she now brings that same attention to detail and charm to helping players. She’s a morning person who swears by Cuban coffee, loves salsa dancing, and is passionate about creating smooth, stress-free experiences for everyone she helps. Her favorite casino game is Lightning Roulette—she loves the suspense and the thrill when the multipliers hit! Why backstories matter:
  • Creates a more human, relatable agent
  • Gives players someone to connect with emotionally
  • Reinforces your brand personality through the agent’s character
  • Makes conversations feel more natural and less transactional
Backgrounds are optional but highly recommended for brands that want to create deeper player engagement. Keep backgrounds appropriate for your audience and aligned with your brand voice.

​
Communication Style Settings

Beyond personality attributes, you can fine-tune how your agent communicates with additional style controls:
Control whether your agent uses emojis in responses. Some brands prefer a playful spirit with emojis (😊 👍 🎉), while others maintain a strictly professional tone without them. This is easily toggled in your agent settings.When to use emojis:
  • Casual gaming brands with younger demographics
  • Social casino environments
  • Promotional or celebratory messages
When to avoid emojis:
  • Premium/VIP brands
  • Formal compliance communications
  • Serious player issues or complaints
Control how your agent acknowledges player messages before processing:Neutral: Simple, professional acknowledgments like “Let me check on that for you.”Empathetic: Warmer acknowledgments that validate player feelings, such as “I totally understand your frustration. Let me look into this right away.”Choose empathetic mode for brands that prioritize emotional connection, or neutral for more streamlined interactions.
Character Substitutions are configured separately under Settings → Language & Formatting, not in agent settings. This feature allows you to find and replace specific characters or punctuation across all AI responses (e.g., replace em-dashes with commas, remove exclamation points, etc.).

​
Personality Attributes

Attributes help guide how the agent interacts with players and handles conversations. Some agents may need a professional tone, while others can be more playful or warm. Consider the type of experience you want your players to have and choose attributes accordingly. Available Personality Attributes:
Understanding and compassionate. The agent shows genuine care for player concerns and frustrations. Best for handling sensitive issues like account problems or responsible gaming scenarios.
Kind and inviting. Creates a friendly, welcoming atmosphere in every interaction. Great for building player loyalty and comfort.
Reliable and expert. Maintains a polished, business-like demeanor while remaining helpful. Ideal for premium brands or VIP player support.
Lighthearted and playful. Incorporates appropriate humor to keep conversations engaging. Best for casual gaming brands with younger demographics.
Formal and focused. Maintains professionalism and stays on-topic. Useful for compliance-related communications or high-stakes issues.
Clear and efficient. Gets straight to the point without unnecessary elaboration. Ideal for players who prefer quick, direct answers.
Encouraging and positive. Frames responses in an uplifting way while still being honest. Helps maintain player morale.
Helpful and advisory. Takes a teaching approach to help players understand processes and solutions.
Solution-oriented and focused on outcomes. Prioritizes resolving the issue efficiently over lengthy explanations.
Thoughtful and attentive. Carefully considers player circumstances and context before responding.
Inspiring confidence and enthusiasm. Encourages players and celebrates their successes.
Adaptable and understanding. More willing to find creative solutions and make exceptions when appropriate.
Straightforward and no-nonsense. Eliminates fluff and focuses on essential information only.
Playful and charming. Uses subtle, appropriate charm to create engaging interactions. Use carefully and only for brands where this aligns with player expectations.
Pro Tip: Combine 2-4 attributes to create a nuanced personality. For example, “Empathetic + Professional + Concise” creates a caring but efficient agent perfect for VIP support.

​
Reply Examples

A powerful way to train your agent’s tone is through reply examples. These are sample responses (1-10 examples) that demonstrate exactly how you want your agent to sound. How it works:
  • Provide real examples of how your agent should respond in various scenarios
  • The AI learns your preferred phrasing, vocabulary, and style patterns
  • More examples = better tone consistency
Example Reply Samples for a Friendly Gaming Brand:
  • “Hey! Happy to help with that withdrawal question 😊”
  • “I totally understand the frustration—let me look into this for you.”
  • “Great news! I found the information you need.”
  • “No worries at all! That happens sometimes. Let’s get it sorted.”
Example Reply Samples for a Premium/VIP Brand:
  • “I’d be delighted to assist you with your account inquiry.”
  • “Thank you for your patience. I’ve reviewed your withdrawal request.”
  • “I understand your concern. Allow me to investigate this matter for you.”
  • “Your account has been updated as requested.”
Start with 3-5 reply examples that capture your brand’s voice, then add more as you refine your agent’s personality.

​
Agent Instructions

This is where the magic happens. Think of agent instructions as sending out a company-wide memo to all your human agents with clear guidelines on how to handle tickets. These instructions dictate how the AI agent responds, ensuring alignment with your brand’s voice and policies. What to include in agent instructions:
  • Brand-specific terminology and phrasing preferences
  • Current promotions or offers to mention
  • Consistent sign-offs or signatures
  • Escalation triggers and when to involve human agents
  • Tone guidelines for specific scenarios

​
Example Use Cases

  • Ongoing Promotions: “Always mention our Weekend Reload Bonus when players ask about deposit bonuses or if denying goodwill bonus requests.”
  • Consistent Sign-Offs: “Sign off every message with ‘Happy Gaming! - [Agent Name]’”
  • VIP Treatment: “For VIP players, always use formal language and offer priority processing when available.”
  • Responsible Gaming: Handled automatically by the platform — configure in Compliance Settings and Content Shield

​
How Your Instructions Work with Platform Defaults

Cevro comes with built-in tone and style guidelines that ensure every agent communicates clearly, warmly, and professionally out of the box. When you write custom instructions for your agent, the system automatically combines your rules with these platform defaults. Your instructions always take priority. If your custom instructions address the same topic as a platform default, your version wins. For example, if the platform default suggests using emojis and your instructions say “never use emojis,” the agent will follow your rule. When there’s no conflict, both sets of guidelines work together. Your brand-specific rules sit alongside the platform’s best practices for things like response structure, escalation etiquette, and professional tone.
Changes to your instructions are processed automatically when you save. There may be a short delay (about 15 seconds) while the system merges your updates with the platform defaults.

​
Writing Effective Agent Instructions

The best agent instructions are specific, clear, and actionable. Think of them as rules your agent follows literally. Be specific, not vague:
Instead of…Write…
”Be friendly""Use contractions (I’ll, you’re, let’s). Start replies by addressing the issue, not with ‘Thank you for contacting us.’"
"Sound professional""Use full sentences and proper grammar. Address players as ‘Sir’ or ‘Madam’ unless they provide a surname."
"Mention promotions""When a player asks about deposit bonuses, mention the Weekend Reload Bonus (50% match up to $200)”
Use firm, positive language. Words like “Always” help reinforce expected behavior. Prefer telling the agent what TO do rather than what NOT to do — positive targets are easier for AI to follow:
  • “Always sign off with ‘Best regards, [Agent Name]’”
  • “When discussing timelines, say ‘I’ll look into this and get back to you’ rather than giving specific hours or days”
  • “Always acknowledge the player’s concern before providing a solution”
Include terminology rules when your brand uses specific language:
  • “Say ‘purchase’ instead of ‘deposit’ and ‘redemption’ instead of ‘withdrawal’”
  • “Refer to currency as ‘Gold Coins’ and ‘Sweeps Coins,’ never as ‘money’ or ‘cash’”
  • “Use ‘play’ and ‘spin’ instead of ‘bet’ or ‘wager’”
Set conditional behaviors for different player segments:
  • “For VIP players (tier 3 and above), use formal language and offer priority processing”
  • “When a player has been waiting more than 5 minutes, acknowledge the wait and apologize”
Common pitfalls to avoid:
  • Repeating what’s already built in — You don’t need to say “be warm and professional” because the platform defaults already cover this.
  • Contradicting AIP instructions — Agent instructions control tone and style. Specific workflow logic (like when to escalate or which tools to use) belongs in your AIPs, not in agent instructions.
  • Being too vague — “Be helpful” gives the agent nothing concrete to act on. Specific rules produce consistent behavior.

​
Brand-Specific Agent Examples

Different brands require different agent personalities. Here are real-world examples of how to configure agents for various brand types:

​
Casual Gaming Brand (18-35 demographic)

Personality Attributes: Funny, Warm, Optimistic Emoji Usage: Enabled ✅ Message Style: Concise Acknowledgment Mode: Empathetic Reply Examples:
  • “Hey there! 😊 Let me help you with that bonus!”
  • “Oh no! That’s frustrating. Let me check what’s going on with your account.”
  • “Awesome! Your withdrawal is on its way! 🎉”
  • “No worries at all—these things happen. Let’s get it sorted!”
Agent Instructions: 
- Use contractions (I'll, you're, let's) and short sentences. One idea per message.
- When a player mentions a win or milestone, acknowledge it: "Nice hit! 🎉"
- Sign off every conversation with "Happy Gaming! 🎮"
- Use the player's name only in the greeting. After that, address as "you."
- Start replies by addressing the issue directly — skip "Thank you for reaching out."
Result: An approachable, fun agent that makes players feel comfortable and engaged.

​
Premium/VIP Brand (High-value players)

Personality Attributes: Professional, Empathetic, Considerate, Warm Emoji Usage: Disabled ❌ Message Style: Detailed responses Acknowledgment Mode: Neutral Reply Examples:
  • “I’d be delighted to assist you with your withdrawal inquiry.”
  • “Thank you for your patience while I review your account.”
  • “I understand your concern. Allow me to investigate this matter immediately.”
  • “Your request has been processed with priority status.”
Agent Instructions: 
- Use full sentences and proper grammar. Address players as "Sir" or "Madam" unless they provide a surname.
- Use the player's name only in the greeting and when closing. After that, address as "you."
- When a VIP player (tier 3+) contacts support, acknowledge their status: "As a valued VIP member, I'll prioritize this for you."
- Sign off with "Best regards, [Agent Name]"
- Keep responses thorough but structured — lead with the answer, then provide supporting details.
Result: A refined, sophisticated agent that makes high-value players feel valued and respected.

​
Sports Betting Brand (Action-focused, competitive players)

Personality Attributes: Results-Driven, Concise, Motivational, Streamlined Emoji Usage: Limited (only for celebrations) Message Style: Very concise, direct Acknowledgment Mode: Neutral Reply Examples:
  • “Checking your bet slip now.”
  • “Found it. Your payout is processing—arrives in 24 hours.”
  • “Good catch. I’ve corrected the odds on your account.”
  • “You’re all set. Good luck! 🍀”
Agent Instructions: 
- Keep replies to 1-2 sentences. One fact per message.
- Lead with the answer, then provide context only if needed.
- When a player wins a bet, acknowledge briefly: "Nice call." or "Great pick."
- Skip opening pleasantries — start with the information the player asked for.
- Use the player's name only in the greeting.
Result: A fast, efficient agent that respects players’ time and matches their competitive, action-oriented mindset.

​
Responsible Gaming — How Cevro Handles It

Responsible gaming (RG) detection is built into the platform — you don’t need to write agent instructions for it. Cevro has two layers of protection that work automatically: Layer 1: Compliance Settings — Built-in RG detection that runs on every player message. When a player expresses gambling-related distress (addiction, financial hardship, self-exclusion requests, emotional distress), the system detects it and responds based on the action mode you configure:
  • Escalate — Hand off to a human agent with a supportive message
  • Empathic Response — AI responds with empathy and resources, no handoff
  • Follow Instructions — Define custom handling with the ability to route to specialized AIPs
Layer 2: Content Shield — A pre-AI safety net for the most critical content (self-harm, suicidal intent). When triggered, zero automated messages reach the player — only a human communicates. Configure this through Automation Rules.
RG detection is always on — there is no toggle to disable it. Content Shield runs on every inbound message, catching concerns even mid-conversation after the player has been routed into an AIP flow.
What you DO configure in agent instructions: You don’t need to tell the agent how to detect RG concerns — the platform handles that. But you can customize the tone of the response through your Compliance Settings. For example, setting an example response that matches your brand voice: 
"I hear you, and I want you to know your wellbeing comes first.
I'm connecting you with a specialist who can help. In the meantime,
you can reach the 24/7 helpline at 1-800-GAMBLER."
Configure your RG action mode and example response in Settings → Agent Configuration → Compliance. For self-harm detection via Content Shield, set up a Content Detected automation rule in Settings → Automation Rules. See the Content Shield guide for setup steps.

​
Instruction Writing Cheat Sheet

Quick reference for writing instructions that work reliably:
GoalWrite this (specific, positive)Not this (vague or negative)
Less name repetition“Use the player’s name only in the greeting. After that, address as ‘you’.”“Don’t use the player’s name too much”
Shorter responses“Keep replies to 1-2 sentences. One idea per message.”“Don’t write long messages”
More natural tone“Use contractions (I’ll, you’re, let’s). Skip ‘Thank you for contacting us’ — just address the issue.”“Sound more human”
Formal tone“Use full sentences, proper grammar. Address players as ‘Sir’ or ‘Madam’.”“Be professional”
Promote an offer“When a player asks about bonuses, mention the Weekend Reload Bonus: ‘50% match up to $200.’ Only mention if they’re asking about promotions — not when they’re frustrated.”“Mention promotions”
Handle unknowns“When you can’t find the answer, say ‘Let me connect you with a specialist who can help.’”“Don’t say ‘I don’t know’”
Lead with solutions“Start your reply with what the player can do. Put the action step first, explanation second.”“Be solution-oriented”
Stay on topic“If the player asks something unrelated to their account, redirect: ‘I’m best at helping with account questions — what can I look into for you?’”“Stay focused”
The pattern is simple: say what TO do, not what to avoid. AI agents follow positive instructions more reliably than prohibitions. Instead of “never do X,” describe what the agent should do instead.