In the previous video, we learned what Chat IQ is and the key requirements needed to get started.

Now, let’s move to the next step — setting up Chat IQ! In this video, we’ll walk you through the setup process step by step and help you get your Chat IQ Agent live.

Steps:

Step 1: Request Access to Chat IQ

• First, navigate to the AI Suite option from the side navigation bar in telecrm.
• Click on Request Access. You can either copy the email address provided or open it directly in your mail app. Send an access request to the telecrm team.
• Once your request is approved, usually within 24 to 48 hours, you can start setting up your Chat IQ Agent.

Step 2: Create Your First Agent

• After your access is approved, return to the Chat IQ page and click on Setup Chat IQ Agent.
• Next, click on Create One to create your first agent.
• Start by giving your agent a name. For example, we'll call it Initial Enquiries. This makes it easier to identify the purpose of your agent and where it will be used.

Step 3: Select the AI Model and Add Your API Key

• Next, select the AI model you want your agent to use. We recommend GPT-4o mini as a starting option.
• Now, copy your OpenAI API key from your OpenAI account and paste it into the required field.
• This connects your agent to the selected AI model.

Step 4: Define the Agent's Instructions

• The next step is to define the instructions, also known as the prompt, that tell your agent how to respond to customers.
• To help you get started, we've provided a sample prompt:

“You are a helpful sales assistant replying to customers on WhatsApp. Answer using only the conversation and lead context provided. Ask one relevant question at a time, keep replies clear and concise, and never invent information.”

• You can modify these instructions according to your business requirements and the type of conversations you want your agent to handle.

Step 5: Choose the Agent's Tone

• Now, let's define how your agent should communicate with customers.
• You can choose from different tones, including Warm, Respectful, Neutral, Firm, Playful, Formal, and Reassuring.
• You can select multiple tones depending on the kind of communication you want your agent to maintain with your customers.

Step 6: Set the Reply Length

• Next, decide how long your agent's messages should be while responding to customers on WhatsApp.
• You can choose between Short, Medium, and Long, depending on your use case.
• As a starting point, we recommend choosing Medium, with a reply length of around 700 characters. This helps your agent provide useful information without making every response unnecessarily lengthy.

Step 7: Configure the Response Format

• The next section is the Response Format. This defines how the agent's responses will be structured and what information it should return.

Let's understand the default fields one by one.

• Reply: Contains the response the agent wants to send to the customer.
• Handoff Requested: Indicates whether the agent needs to hand over the conversation to a human team member. When the handoff rules are triggered, this field is set to “Yes,” and the agent stops replying until the conversation is handed back.
• Handoff Reason: Explains why the agent has requested human intervention.
• Reasoning: Captures the agent's reasoning output, as configured by the system.
• These fields are locked because they are part of the system's default response structure and are required for the agent to function correctly.
• You can also add custom fields to collect additional information during a conversation. This step is optional, but it can be useful for your business.
• For example, a coaching institute may offer multiple batches, such as JEE 2-Year, JEE Dropper, and NEET. The institute can add an Interested Plan field with these options. Based on the conversation, the agent can identify the customer's preferred batch and return that information through the configured response or custom action.
• This helps your team capture useful lead information without having to ask for it again manually.

Step 8: Set Up Advanced Handoff Rules

• Now, let's move to the Advanced section.
• Here, you can define the conditions under which the agent should stop responding and hand over the conversation to a human team member.

For example, you can use a rule like:

“Hand off when the customer asks for a human, raises an issue you cannot resolve from the provided context, or requests a commitment you are not authorized to make.”

• Once a handoff rule is triggered, the agent stops responding, and the conversation is assigned to the lead owner as per the handoff workflow. The handoff reason is also recorded so the team knows why the conversation was transferred.

Step 9: Handle Handoffs and Handbacks

• Your team can easily track these conversations in the Handoff section of the WhatsApp conversation window.
• Before replying, the team member can review the conversation history to understand what the customer and the agent have discussed so far.
• Once the team member has resolved the query or completed their part of the conversation, they can click on Handback to return the conversation to the Chat IQ Agent.
• The agent can then resume replying according to its instructions and the conversation context.
• This allows your team to step in whenever human assistance is needed and hand the conversation back when the agent can continue.

Step 10: Configure the Lead Context

• Next, let's look at one of the most important settings: Lead Context.
• Here, you can select the lead fields and activity types that your agent is allowed to read. This information helps the agent understand the customer better and provide more relevant responses.
• For example, suppose your team has recorded payment details, call activities, or custom activities containing information about a special discount promised to a lead.
• If you provide the agent access to these relevant fields and activities, it can use that information when responding to the customer's questions.
• This helps the agent give more informed and context-aware replies instead of relying only on the current WhatsApp conversation.
• Remember to select only the lead fields and activities that are relevant to your agent's purpose.

Step 11: Configure Conversation History

• Next, let's look at the Conversation History setting.
• This setting allows you to decide how many of the most recent WhatsApp messages the agent can read before generating a reply.
• We recommend starting with a limit of 10 messages. Reading more messages can increase token consumption, which may increase your usage costs.
• If your use case requires the agent to consider a longer conversation history, you can increase the limit accordingly. Otherwise, keeping it at 10 is a good starting point.

Step 12: Set Up the Knowledge Base

• Now, let's move to the Knowledge Base section.
• Sometimes, you may have additional information that cannot be included in the agent's main instructions. The Knowledge Base allows you to add this information so your Chat IQ Agent can refer to it while responding to customer queries.
• There are two ways to add knowledge to your Knowledge Base: Manual Entry and File Upload.

Option 1: Add Knowledge Manually

This option is useful when you want to add small pieces of information, such as a paragraph, a frequently asked question, or a short article, to update your agent's knowledge.

There are two types of manual entries you can create: FAQ and OpenText.

• FAQ: Use this option when you want to add a specific question and its answer. For example, you can add a question about your course fees along with the answer you want your agent to provide. This helps the agent answer similar questions from customers.
• OpenText: Use this option when you want to add more detailed information in an article-like format. For example, you can add details about your admission process, company policies, or a particular service you offer.

Option 2: Upload Files

If you have important business information already available in documents, you can upload those files directly to the Knowledge Base instead of adding everything manually.

telecrm supports file uploads in formats such as PDF, TXT, and MD. Please remember that the maximum file size allowed is 10 MB per file.

• Assign Categories to Your Knowledge
• Whether you're adding knowledge manually or uploading files, organising your information into categories helps you manage it more effectively.
• For example, you can create categories such as Sales, Marketing, and Customer Support to keep different types of information separate.
• During file upload, you can select the relevant category for that file. You can then assign the required categories to a particular Chat IQ Agent so it can access the knowledge available under those categories.
• For example, if you have a sales agent, you can assign it the Sales category so it can refer to relevant product details, pricing information, or sales policies while answering customer queries.
• This way, you can control which knowledge is made available to each agent and keep its responses relevant to its purpose.

Understand Knowledge Base File Statuses

On the Knowledge Base page, you'll see different statuses that tell you what's happening with each uploaded file.

Let's understand what each status means.

• Processing: When a file is uploaded, it may show as Processing. This means OpenAI is processing and indexing the file so the agent can retrieve information from it.
• In Use: Once processing is complete, the file is marked as In Use. This means the file is ready to be used by an agent that has access to its assigned category.
• Paused: If a file contains outdated information or you no longer want the agent to refer to it, you can pause it. The file remains in your Knowledge Base, but the agent will no longer use it as a source while it is paused.
• Failed: If a file fails to process, it may be due to an issue with the API key or another processing error. The file won't be available for the agent to use until the issue is resolved and it is processed successfully.
• Active: This status indicates that the knowledge entry or file is active in the system. Its availability to an agent depends on its processing state, category assignment, and configuration.

These statuses help you track your knowledge files and manage the information your agent can use.

Step 13: Set Up Your Workflow Automation

• Once you've configured your agent, the next step is to set up a workflow around it.
• You need to create a flowchart or workflow automation that defines how the Chat IQ Agent will be triggered and how the conversation will be handled.
• Once your workflow is created, activated, and live, its name will be available for reference in the relevant configuration.
• Make sure your workflow is set up according to your business requirements so your agent can participate in the intended conversation flow.

Step 14: Use Debug Mode to Test Your Agent

• Before putting your agent into regular use, you may want to understand why it generates a particular response. This is where Debug Mode comes in.
• When you enable Debug Mode, an additional locked field is added to the Response Format. This field is also available through the custom action, allowing you to review the agent's reasoning output alongside its replies.
• For example, if your agent gives an unexpected answer, Debug Mode can help you understand what information or context contributed to that response.
• This is especially useful when testing your agent, refining its instructions, and improving its behavior.
• However, Debug Mode can increase token consumption. So, we recommend keeping it enabled during testing and turning it off once you're satisfied with the agent's behavior.

And that's how you set up your Chat IQ Agent in telecrm!

In the next video, we'll explore how to create Workflows around Chat IQ.