Overview

The charles AI FAQ Agent allows you to automate answers to your most common customer questions, directly inside WhatsApp. Instead of routing every inquiry to a human agent, the FAQ Agent handles repetitive questions instantly, 24/7, so your team can focus on conversations that truly need a human touch.

Before You Begin

Make sure you have the following ready before starting:

✅ Access to the AI agent tab on the platform
✅ A CSV file containing your FAQs (questions and answers) download a template here
✅ Your brand name, a short company description, and a preferred agent name

Step 1 — Set Up Your AI Agents (Global Configuration)

Before creating any individual agent, you need to complete the global AI setup. These settings act as the foundation for all agents you create.

How to do it:

Open the Charles platform and click "Setup your AI Agents" on the welcome screen.
This opens the global configuration panel where you define settings that every agent will inherit.

💡 Important: Always complete this step first. Skipping it means your individual agents won't have a consistent baseline configuration.

Step 2 — Define Your AI Persona

The AI Persona determines how your agent communicates: its name, tone, and language behaviour across all customer conversations.

How to do it:

Navigate to Charles AI Squad > AI Persona.
Fill in the following fields:

Field	Description
AI Name	The name your agent will use in WhatsApp conversations (e.g., "Ava")
Brand Name	The correct spelling of your brand name
Company Description	A brief overview of your business, products, and services
Fallback Message	The default message shown when the agent cannot understand a customer's request

💡 Pro Tip: Think of the Company Description as a briefing document for a new team member. The more specific and detailed it is, the more consistent and on-brand your agent's responses will be.

📖 Here is an example text for Company description:

C-fashion is an online fashion platform, offering a curated selection of clothing and accessories, such as shirts, dresses, hats, jewellery, and much more.

You will also see other AI persona options: Tone and brand voice, Content and jargon:

You will be able to test your agent so do not worry about getting the tone or content vocabulary perfectly right during setup.

Do not forget to set-up a fallback message:

This message will trigger when no agent is able to answer a user’s question.

Step 3 — Upload Your Knowledge Base

The Knowledge Base is the single source of truth for your FAQ Agent. It contains all the questions and answers the agent draws from when responding to customers.

Prepare Your CSV File

Your CSV file can contain up to 4 columns:

Column	Description
Topic	The question or topic the agent should recognise
Answer	The response the agent will send to the customer
Image URL(optional)	A link to an image to accompany the answer
URL(optional)	A link to a relevant page (e.g., product page, blog post)

💡 Pro Tip: Write your answers in a conversational tone — the agent will use them as-is. Keep them concise and friendly, as if a customer service rep were typing them.

Upload the CSV

Navigate to the Knowledge Base section in the Charles platform. Download a template here.
Upload your CSV file and click Next

3. Preview the upload and map the fields to ensure they align with the Charles data schema.

4. Verify that all entries are complete — empty fields may cause the agent to return incomplete or missing answers.

🛠️ Troubleshooting: Google Sheets CSV Export Issues

If you're exporting your CSV from Google Sheets, newline characters (\n) inside cells can make the file unreadable. Follow these steps to fix it before uploading:

Open your Google Sheet and press Command + F, then click the three dots (⋮) to expand the Find & Replace options.
In the Find field, enter: \n+
In the Replace with field, enter a single space.
Check the option "Search using regular expressions."
Click "Replace all."

Your file is now ready to upload cleanly.

Step 4 — Create Your FAQ Agent

Now it's time to create the individual FAQ Agent and connect it to your Knowledge Base.

How to do it:

Navigate to Charles AI Squad > Agents and click "Create AI Agent".
Select "FAQ Automation" as the agent type.
Give your agent a recognisable name, e.g., "Your Brand Name FAQ Agent".
Under the Knowledge Base tab, select your uploaded CSV from the dropdown menu.

5. Add additional instructions to fine-tune the agent's behaviour. Examples:

"Always reply in the language the customer used."
"If you don't know the answer, say so honestly and offer to connect the user to a human agent or add a support email/phone number to call.

💡 Pro Tip: Additional instructions are a powerful way to handle edge cases and set clear expectations for how the agent should behave when it's uncertain.

Step 5 — Test Your Agent

Before going live, always test your agent to make sure it responds accurately and handles a variety of scenarios correctly.

How to do it:

Navigate to Charles AI Squad > Agent Test.
Select your FAQ Agent by checking the box next to its name under "Select Agents".
Toggle "One Agent Active" to activate the test environment.
Type in queries your customers would typically ask and verify the responses.
Use "Reset Conversation" to clear the chat and start a fresh test scenario.

What to test:

✅ Common questions your customers ask most frequently
✅ Questions the agent should not be able to answer (to verify the fallback message)
✅ Queries in different languages (if multilingual support is expected)
✅ Ambiguous or vague questions to see how the agent handles uncertainty

Step 6 — Enable Your Agent on WhatsApp

Once testing is complete, it's time to go live. You have two deployment options:

Table

Option	Description	Best For
Global AI Agent	The agent handles all incoming WhatsApp conversations by default, replacing the catch-all flow	Brands ready for full automation
Journey Flow Agent	The agent is activated at a specific point within a Journey Flow (e.g., after a customer selects a menu option)	Brands wanting a controlled rollout

💡 Recommendation: If this is your first time deploying an AI agent, start with a Journey Flow deployment. This lets you test the agent with a subset of real conversations before rolling it out globally.

After going live:

Monitor your agent's performance in the Analytics tab at the agent level.

Track key metrics such as resolution rate, escalation rate, and customer satisfaction to continuously improve performance. You're All Set! 🎉

Your AI FAQ Agent is now configured, tested, and live on WhatsApp. Here's a quick recap of what you've completed: