AI Agents — Video Tutorial Script
Video Info
- Duration: ~8 minutes
- Target Audience: Users who want to set up AI-powered customer support and sales assistants
- Prerequisites: Logged into DOLU.ai, at least one channel connected
Script
Scene 1: Introduction (0:00-0:20)
Narrator: "Welcome back to DOLU.ai. In this video, you will learn how to create and configure AI agents — the intelligent assistants that can handle customer conversations, answer questions, and escalate complex issues to your team automatically. Let's get started." Screen: DOLU.ai logo and intro animation displayed
Scene 2: Accessing AI Agents (0:20-0:50)
Narrator: "Click 'AI Agents' in the left sidebar. You will see the AI Agents dashboard showing all your configured agents and their performance metrics." Screen: AI Agents page loads — dashboard shows agent cards with status indicators Action: Click "AI Agents" in the left sidebar — page loads
Narrator: "Each agent card displays the agent's name, active channels, total conversations handled, and the satisfaction rating from customer feedback." Screen: Agent cards with metrics visible — status badges show "Active" in green Action: Hover over agent cards to show summary statistics
Scene 3: Creating a New AI Agent (0:50-2:00)
Narrator: "To create a new agent, click 'Create Agent' in the top-right corner." Screen: "Create Agent" button highlighted Action: Click "Create Agent" — agent creation wizard opens
Narrator: "Step one: give your agent a name and description. The name is what your team sees internally — for example, 'Support Bot' or 'Sales Assistant'. The description helps team members understand what this agent does." Screen: Agent creation form — Name and Description fields Action: Type "Customer Support Agent" in Name, type "Handles general support inquiries and FAQs" in Description
Narrator: "Step two: select the channels this agent should operate on. You can assign it to WhatsApp, Instagram, Telegram, email, your web widget, or any combination." Screen: Channel selection with checkboxes — icons for each channel Action: Check WhatsApp, Web Widget, and Telegram — leave others unchecked
Narrator: "Step three: choose the agent's primary language. This determines the language the agent uses when responding to customers. DOLU.ai supports 14 languages, and the agent will respond in the selected language by default." Screen: Language dropdown showing all 14 supported languages Action: Select "English" from the dropdown
Narrator: "Click 'Create' to set up the agent. We will configure its personality and knowledge base next." Screen: Click "Create" — agent profile page opens Action: Click Create button — agent profile loads
Scene 4: Knowledge Base Setup (2:00-3:20)
Narrator: "An AI agent is only as good as the knowledge you give it. Click the 'Knowledge Base' tab on the agent's profile page." Screen: Agent profile page — "Knowledge Base" tab highlighted Action: Click "Knowledge Base" tab — knowledge management interface opens
Narrator: "Here you can add the information your agent needs to answer customer questions. Click 'Add Knowledge' and choose a category: FAQ, Product Info, Policy, or General." Screen: "Add Knowledge" button — category selector dropdown Action: Click "Add Knowledge" — select "FAQ" category
Narrator: "Enter the question and the answer. For example — Question: 'What are your business hours?' Answer: 'We are available Monday through Friday, 9 AM to 6 PM Eastern Time. You can also reach our AI assistant 24/7.'" Screen: Knowledge entry form with Question and Answer fields Action: Type the question and answer — click Save
Narrator: "Add as many entries as you need. The more knowledge you provide, the more accurately your agent can respond. You can also import knowledge in bulk from a JSON or CSV file." Screen: Click "Import" button — file upload area appears Action: Show the import option — then close and continue adding entries manually
Narrator: "Pro tip: review your most common customer questions from the inbox and add them here first. This immediately covers the majority of incoming inquiries." Screen: Text overlay: "Start with your top 20 most frequently asked questions"
Scene 5: Personality Configuration (3:20-4:20)
Narrator: "Click the 'Personality' tab to define how your agent communicates. This is what makes your AI agent feel like a real part of your brand." Screen: Personality tab opens with configuration options Action: Click "Personality" tab — settings load
Narrator: "First, set the tone. Choose from Professional, Friendly, Casual, or Formal. This affects word choice, sentence structure, and how the agent greets customers." Screen: Tone selector with four options — visual examples of each tone shown Action: Select "Friendly" — preview text updates to show a sample response in friendly tone
Narrator: "Next, define the greeting message. This is the first thing the agent says when a new conversation starts. For example: 'Hi there! Welcome to Acme Support. How can I help you today?'" Screen: Greeting message text field Action: Type the greeting message — preview updates in real time
Narrator: "You can also set the agent's name as it appears to customers. This can be different from the internal name. For example, internally it is 'Customer Support Agent' but customers see 'Alex from Acme'." Screen: Display name field — type customer-facing name Action: Type "Alex from Acme" — preview shows how it appears in the chat widget
Narrator: "Finally, add custom instructions. These are specific rules for your agent to follow — for example, 'Always suggest scheduling a demo for enterprise inquiries' or 'Never share competitor pricing'." Screen: Custom instructions text area Action: Type "Always suggest scheduling a demo for enterprise inquiries" — click Save
Scene 6: Auto-Reply Toggle (4:20-5:00)
Narrator: "The auto-reply toggle controls whether the agent responds to incoming messages automatically. When enabled, every new message on the assigned channels triggers an AI response." Screen: Auto-reply toggle switch — currently set to Off Action: Click to toggle On — confirmation dialog appears
Narrator: "You can set auto-reply to operate on a schedule. For example, enable auto-reply only outside business hours — between 6 PM and 9 AM — so your team handles inquiries during the day and the AI covers the rest." Screen: Schedule configuration — time range selectors Action: Set start time to 18:00 — set end time to 09:00 — select all days — click Save
Narrator: "You can also set a response delay to make the interaction feel more natural. A 2-3 second delay before responding prevents the 'instant bot' feeling." Screen: Response delay slider — set to 3 seconds Action: Adjust the delay slider from 0 to 3 seconds — click Save
Scene 7: Feedback Loop — Thumbs Up and Down (5:00-5:50)
Narrator: "DOLU.ai includes a built-in feedback loop to help your AI agent improve over time. After each AI response, customers can click a thumbs up or thumbs down icon." Screen: Chat window showing AI response with thumbs up/down icons below it Action: Highlight the feedback icons
Narrator: "Positive feedback confirms the response was helpful. Negative feedback flags the response for review. Go to the 'Feedback' tab on the agent's profile to see all feedback." Screen: Feedback tab — list of responses with thumbs up (green) and thumbs down (red) indicators Action: Click "Feedback" tab — scroll through feedback entries
Narrator: "For each negative feedback entry, you can see the original question, the AI's response, and the customer's reaction. Click 'Correct' to provide the right answer." Screen: Click on a negative feedback entry — correction form opens Action: Click "Correct" — type an improved response — click Save
Narrator: "Once you save the correction, the AI agent learns from it immediately. The next time a similar question comes in, it will use the corrected response. This is how your AI gets smarter over time." Screen: Confirmation message: "Correction saved — AI knowledge updated" Action: Observe the confirmation — entry status changes from red to green
Scene 8: Training from Corrections (5:50-6:30)
Narrator: "Beyond individual corrections, your team members can train the AI directly from the inbox. When a human agent sees an incorrect AI response, they can click the 'Train AI' button next to the message." Screen: Inbox conversation — AI response shown — "Train AI" button highlighted next to the message Action: Click "Train AI" — training form opens within the conversation
Narrator: "The training form shows the customer's question and the AI's response. Edit the response to what it should have been, add an explanation, and submit." Screen: Training form with original and corrected response fields Action: Edit the response — type a better answer — click Submit
Narrator: "All training submissions appear in the 'Training Queue' on the AI Agents page. An admin can review and approve or reject each submission before it is applied." Screen: Training Queue tab — list of pending submissions with Approve/Reject buttons Action: Click Approve on a submission — entry moves to "Applied" status
Scene 9: Escalation Rules (6:30-7:40)
Narrator: "Not every conversation should be handled by AI. Escalation rules define when the agent should transfer a conversation to a human team member." Screen: Escalation Rules tab on the agent's profile page Action: Click "Escalation Rules" tab — rule configuration interface loads
Narrator: "DOLU.ai comes with default escalation triggers: when the customer explicitly asks for a human, when the AI's confidence score falls below a threshold, or when the customer expresses strong frustration." Screen: Default rules listed with toggle switches Action: Scroll through default rules — all enabled
Narrator: "You can add custom escalation rules. Click 'Add Rule' and define the trigger. For example, 'Escalate when the customer mentions refund or cancellation'." Screen: "Add Rule" button — rule creation form Action: Click "Add Rule" — enter keyword trigger "refund OR cancellation" — set action to "Assign to Support Team" — click Save
Narrator: "When an escalation triggers, the conversation is assigned to the specified team or team member, and a notification is sent immediately. The human agent sees the full conversation history including AI responses." Screen: Example notification: "AI escalated: Customer requesting refund — Assigned to Support Team" Action: Show the notification in the notification panel
Narrator: "You can also set a maximum number of AI exchanges before automatic escalation. If the conversation goes back and forth more than, say, five times without resolution, it escalates automatically." Screen: "Max exchanges before escalation" setting — numeric input Action: Set value to 5 — click Save
Scene 10: Closing (7:40-8:00)
Narrator: "You now know how to create, configure, and continuously improve your AI agents in DOLU.ai. From knowledge setup and personality tuning to feedback loops and escalation rules, your AI agent is ready to serve your customers around the clock. In the next video, we will deep dive into the knowledge base. See you there!" Screen: DOLU.ai logo with text overlay: "Next Video: Knowledge Base"
Summary
What you learned in this video:
- How to create a new AI agent and assign it to channels
- How to populate the knowledge base with FAQs, product info, and policies
- How to configure the agent's personality, tone, and greeting
- How to manage the auto-reply toggle and scheduling
- How the feedback loop (thumbs up/down) drives continuous improvement
- How to correct AI responses and train the agent from the inbox
- How to set up escalation rules for human handoff
- How to use the training queue for team-based AI improvement
Next Video
Knowledge Base — Manage the information that powers your AI agents.