About this video

• Video Title: AI AUTOMATION TECH BOOT CAMP DAY 4
• Channel: Jokotoye Emmanuel | Automation King
• Speakers: Jokotoye Emmanuel
• Duration: 02:47:37

Overview

This video is the fourth day of an AI Automation Tech Boot Camp. The speaker, Jokotoye Emmanuel, emphasizes the crucial role of APIs in automation and provides a detailed explanation of what APIs are, how they work, and their various applications. He also demonstrates how to connect with popular APIs like OpenAI, Google Gemini, and Telegram within an automation platform, and guides viewers through building a practical AI-powered chatbot for a restaurant. The session concludes with an assignment for participants to build their own AI customer support agent.

Key takeaways

• API Fundamentals: APIs act as messengers between applications, allowing them to communicate and share data without revealing their internal complexities.
• API Components: Understanding endpoints (addresses), methods (GET, POST, PUT, DELETE), headers, and bodies is essential for successful API integration.
• Practical API Integration: The video demonstrates connecting with OpenAI, Google Gemini, and Telegram APIs, highlighting the process of obtaining API keys and configuring them within an automation workflow.
• AI Chatbot Development: A practical example of building an AI customer support chatbot for a restaurant (Chef Caps) is shown, utilizing an AI agent, knowledge base, and various tools.
• Assignment: Participants are tasked with creating their own AI customer support agent for a business, documenting their process, and sharing it on Twitter.

The practical example demonstrated was building an AI-powered chatbot for a restaurant called "Chef Caps."

Here's an elaboration of the steps taken on the screen:

1. Setting up the AI Agent:

   • A chat trigger was used to start the workflow.
   • An AI agent was connected to this trigger.
   • This AI agent was configured with tools like memory and a chat model (using OpenAI's GPT-4).

2. Creating a Knowledge Base:

   • The speaker used ChatGPT (specifically GPT-4) to act as the founder of Chef Caps restaurant.
   • The request was to create a comprehensive FAQ and knowledge base for the restaurant, covering details like location, opening hours, menu, pricing, social media handles, and ordering channels.
   • The output from ChatGPT, detailing Chef Caps' information, was saved.

3. Integrating the Knowledge Base:

   • The generated information was to be uploaded to a Google Doc to serve as the knowledge base.
   • The speaker showed the process of getting the Google Docs API credentials from the Google Cloud Console. This involved:
     • Creating a new project (e.g., "AI Mastery Academy").
     • Enabling the Google Docs API.
     • Creating API credentials (Client ID and Client Secret).
     • Adding the speaker's email as a test user in the Google Cloud Console's OAuth consent screen.
   • These credentials (Client ID and Client Secret) were then entered into the N8N platform to connect the Google Docs tool.
   • The URL of the Google Doc containing the restaurant's information was then used in the N8N workflow to configure the "Get Documents" node, effectively making it the knowledge base for the AI agent.

4. Configuring the AI Agent's Prompt:

   • The speaker asked ChatGPT to create a system prompt for the AI agent.
   • This prompt instructed the AI agent to act as "Chef Caps official customer support," be friendly and helpful, answer customer inquiries, collect lead information (name, phone number, email), and utilize the knowledge base tool and a "think" tool for better responses.

5. Connecting Tools:

   • The AI agent was connected to the OpenAI tool (for the chat model) and the memory tool.
   • The Google Docs tool was connected to serve as the knowledge base.
   • The "think" tool was also integrated.

6. Testing the Chatbot:

   • The speaker initiated a conversation by typing "Hello."
   • The AI agent responded by welcoming the user and asking for their contact information (name, phone number, email) to follow up on inquiries.
   • The speaker provided a name, phone number, and email address.
   • The AI agent then responded with the restaurant's location, opening hours, and menu details, drawing information from the connected Google Doc knowledge base.
   • The demonstration showed how the AI agent could handle various customer queries related to the restaurant.

7. Assignment Details:

   • The speaker provided an assignment for participants to build a similar AI customer support agent for a business of their choice.
   • The assignment requires creating a Loom video showcasing the project, including the problem, solution, AI agent capabilities, system prompts, knowledge base, integrated tools, a live demo of conversations, lead collection via a Google Sheet, and the business impact.
   • The submission deadline and platform (Twitter) were also mentioned.