Use Case Examples

Use case examples presents three practical use case examples of AI agents built on Memorly.AI. Below is a brief overview of each examples with different complexities:

Beginner-Level Agent: FAQ Chatbot

Automates responses to frequently asked customer queries.
Uses a predefined knowledge base for instant replies.
Simple setup with minimal customization required.

Example: A customer support chatbot for an e-commerce store that provides answers to common questions such as "What is your return policy?", "How do I track my order?", and "What payment methods do you accept?" The chatbot pulls answers from the knowledge base and provides instant responses.

To create a knowledge base → Open Knowledge Base → Click on "Create" → Add details

Static data in a knowledge base means that the information remains unchanged and does not update over time. In contrast, dynamic data allows real-time updates, meaning new information can be added to the knowledge base using an API.

To create Agent → Open Agents → Click on "Create" → Select Knowledge Base as created earlier → Agent created✅

After selecting knowledge base → Agent is created

Intermediate-Level Agent: Lead Qualification Bot

Handles complex customer support queries using AI.
Integrates with external APIs to provide real-time data.
Utilizes NLP and sentiment analysis for personalized interactions.

Example: A technical support assistant for a SaaS platform that understands complex troubleshooting requests. It integrates with a knowledge base to provide step-by-step solutions, connects to external APIs to check service status, and escalates unresolved issues to a human support agent when necessary.

Create knowledge base

Create Tools

REST is an architectural style for designing networked applications using standard HTTP methods.

JSON-RPC is a remote procedure call (RPC) protocol that allows invoking functions over a network using JSON.

Headers are the type of content we are using in the tool.

GET (Retrieve Data): Used to fetch data from a server.It is idempotent, meaning multiple requests will not change the resource.

POST(Create Data) : Used to send data to the server and create a new resource.Not idempotent (multiple requests create multiple resources).

PUT(Update/Replace Data) : Used to replace an entire resource.Idempotent (multiple requests with the same data produce the same result).

PATCH(Partial Update) : Used to update only specific fields in a resource.Not idempotent (behavior can change with different partial updates).

DELETE(Remove Data) : Used to delete a resource.Idempotent (deleting the same resource multiple times has no additional effect).

CALL : Sometimes used informally in certain APIs or frameworks as an alias for POST or JSON-RPC calls.

uid is present in the agent web chat connect inside the script you will find the account ID(for eg window.accID = '4';)

Body schema depends upon the tool which we are going to create.

Tool Marketplace : Tools can be created by using the marketplace from the platform also it includes

Send Message Template : This tool to send message templates creates in the Memorly.AI platform.

Firstly after creating the message template copy the template ID
Copy the template ID from the message template and paste it in the schema.
Add the tool to any of the agent and open the live chat session and give the promt saying "send me the template".It will give you the template which is created.

Update Google Sheet : This tool allows users to append new rows of data into a specified Google Sheet. With this tool, AI agents can seamlessly update user information, such as names, phone numbers, and other details, by interacting with Google Sheets in real time. It is an ideal solution for managing and storing dynamic data in a collaborative, accessible format.

Take a google sheet and copy the sheet id which is selected in the below image.

Copy the sheet ID and paste it in sheet_id of the body schema.
range_name depends upon the columns we are having.If we are having the E number of columns then the range_name will be "Sheet1!A:E".
How It Works
The tool works by accepting the following parameters:
- sheet_id (String): The unique identifier of the Google Sheet to which the data will be added.
- col_len (Integer): The number of columns in the sheet, which helps in ensuring that the data is aligned correctly.
- values (List): A list of values to be appended to the Google Sheet. Each item in the list represents a new row of data, where each inner list contains the values for each column.
- uid (String): A unique identifier for the operation or the request.

Meta info : The values of the column names and their description.
Prompt : Tell the agent to refer the particular tool for particular action

Phonpe Payment Link : The PhonePe Payment Link Generator API allows businesses to create secure and seamless payment links for their customers. By integrating this API, merchants can generate dynamic payment links that customers can use to complete transactions through PhonePe's payment gateway.

The client sends a POST request to the API endpoint:https://domain/payments/phonepe/link/generate/.
The request includes headers specifying the content type as JSON.
The request body must contain:

amount (Number) – The transaction amount.
phone_number (String) – The customer’s phone number.
api_url (String, default: https://api-preprod.phonepe.com/apis/pg-sandbox/pg/v1/pay) – The payment processing API endpoint.
marchant_id (String) – Merchant’s unique identifier.
salt_key (String) – Security key for authentication.
redirect_url (String) – URL where users are redirected after payment.
callback_url (String) – URL for receiving payment status updates.

The API processes the request and generates a unique payment link. The client receives the payment link, which can be shared with the customer for payment.

Schedule Google Meet : This tool allows users to easily schedule a Google Meet meeting by providing essential meeting details like title, start time, end time, and timezone. This tool integrates seamlessly with your AI agent, enabling automated scheduling of virtual meetings directly on Google Meet.

How It Works

The tool works by accepting the following parameters:

uid (String): Unique identifier for the meeting.
summary (String): The title or summary of the meeting (e.g., "Team Sync").
start_time (String): The start date and time of the meeting in ISO 8601 format (e.g., "2024-11-02T09:00:00").
end_time (String): The end date and time of the meeting in ISO 8601 format (e.g., "2024-11-02T10:00:00").
timezone (String): The timezone of the meeting. Default is "Asia/Kolkata".

Once the user provides these inputs, the tool automatically schedules the meeting using Google Meet's API and returns a link for participants to join. The tool is flexible, accommodating meetings across different time zones and ensuring the meeting is scheduled accurately.

Get Google Sheet Values : This tool helps in getting data from the sheet and send the input to the user according to the data in the sheet.

In this tool also copy the sheet_id of the google sheet.
Copy the sheet ID and paste it in sheet_id of the body schema.
range_name depends upon the columns we are having.If we are having the E number of columns then the range_name will be "Sheet1!A:E".

Create Agent :

Create Agent, after selecting respective knowledge base and tools.

Planner Prompt : A Planner Prompt is a structured instruction that outlines the purpose, tools, execution steps, and constraints of a task or automation process. It acts as a blueprint to ensure clarity, consistency, and efficient use of resources.

Generator Prompt : A Generative Prompt is a structured instruction that guides an AI system to generate step-by-step responses based on user inputs and specific matching criteria. It ensures interactive, logical, and human-like responses while maintaining accuracy and relevance.

After clicking on "inbox", in your agent, you will be redirected to the support app of your agent, where all the interactions with the customers will be available.

Click on 3 dots option in your agent → Click "Chat" → Chatbot active✅

Advanced-Level Agent: AI-Powered Support Assistant

Engages potential customers and categorizes leads based on responses.
Uses conditional logic to determine lead intent.
Integrates with CRM platforms for streamlined lead management.

Example: A real estate chatbot that asks users about their budget, preferred location, and property type. Based on their answers, the bot qualifies leads into categories such as "High Intent Buyer," "Interested Renter," or "Casual Browser," then forwards qualified leads to sales representatives.

Create knowledge base: