Welcome to the Lex Chatbot series!

In this series, you'll create an AI-powered banking assistant! Designed for absolute beginners.

Introduction

Welcome to this series on building a chatbot with Amazon Lex! 🤖

Over the next 5 projects, you'll build a complete banking chatbot with Amazon Lex, picking up practical AI skills that you can add to your portfolio, resume and other personal projects.

More and more businesses are using conversational AI to connect with their customers in new ways, and this series is the perfect place to start building your AI chatbot skills from scratch.

By the end of this challenge, you'll have built a fully functioning AI chatbot that can:

Greet users and understand different ways people ask for help
Recognize specific banking terms and account types
Retrieve account balances using backend logic
Remember user information to create smoother conversations

What to Expect

This is a 100% hands-on challenge. You'll be working in the AWS console from day one!

Don't worry if you're new to AWS or AI - we've designed this challenge to be totally beginner-friendly:

🤝 Step-by-step instructions guide you through each project. You've got this!
🎥 Live demos of every project on YouTube if you prefer to watch and follow along.
📖 Clear explanations of chatbot concepts as we go (no confusing jargon).
📝 Line-by-line explanations of any code (no coding experience needed at all).
🔍 Troubleshooting tips to help you solve common errors (we've all been there!).
🔥 A supportive community to answer your questions and cheer you on.

What do I need to start?

You'll need an AWS account for this challenge. If you haven't created one yet, you can set one up here.

The best part? Amazon Lex has a generous free tier, so you can complete this entire challenge without spending 👏 anything 👏

We've made sure this whole challenge stays well within those limits, so you can focus on building awesome stuff!

Your 5-Day Challenge

Here's your day-by-day plan for the Chatbot Challenge. Each day builds on what you learned before, so by day 5, you'll have something really impressive to show off!

Day 1️⃣: Build Your First Chatbot with Amazon Lex
Day 2️⃣: Add Custom Slots to Your Lex Chatbot
Day 3️⃣: Connect Your Lex Chatbot to Lambda
Day 4️⃣: Save User Info with Your Lex Chatbot
Day 5️⃣: Build Complex Conversations with Multiple Slots

Got questions?

"How long will each day's project take me?"

Each project is designed to be something you can knock out in one 1-hour sitting - some are quicker, others might take a bit longer if you're exploring.

"Will this cost me money?"

Nope! AWS Lex has a generous free tier that we'll stay well within.

"Do I need to know how to code?"

Nope! We'll use a tiny bit of JavaScript with Lambda later in the challenge, but we'll give you all the code you need with clear explanations. No coding experience necessary.

"Can I use what I learn for different kinds of chatbots?"

Absolutely! While we're building a banking bot, the skills transfer perfectly to any industry you're interested in - from helping patients schedule appointments to assisting students with coursework.

If you're up for a bit of a challenge, quiz yourself on the key concepts up ahead in this project.

What are chatbots and Lex anyway?

Wooooo that's a great question! To say thanks for making it to the end of this challenge intro, you've unlocked a bonus section (and bonus documentation) on AI, chatbots and Lex.

Pick the explanation style that matches your learning preference: 👇

First things first

Why are you building a chatbot? Once you write it down, your purpose becomes so clear and is a source of motivation to get this series done!

Some awesome reasons students have shared before are...

"I want to show my technical AI skills to potential employers."
"I'm fascinated by AI and want to learn how it works."
"My company needs to automate customer service, and I want to lead that initiative."
"I want hands-on experience with AWS services beyond just studying for certifications, and AI is a growing AWS offering!"

What is AI?

Artificial Intelligence (AI) is like teaching computers to think and learn in ways similar to humans. Instead of just following exact instructions, AI systems can figure things out on their own based on examples and experience.

AI is a huuuuuuuuge field with many specialized branches – kind of like how "sports" includes everything from swimming to chess. Here are some of the main types:

Natural Language Processing (NLP) - Teaching computers to understand and respond in human language
Generative AI - Creating new content like images, text, or music
Speech recognition - Converting spoken words into text (like when you talk to Siri)
Recommendation systems - Suggesting products or content you might like (based on your activity and preferences)
Computer vision - 'Seeing' and analysing images and videos

Chatbots use several AI technologies to create natural-feeling conversation with users:

NLP to understand what you're asking.
Generative AI to create helpful answers.
Speech recognition to capture input given through speech.

Now that you know about AI and chatbots, you might be wondering...

How do you build a chatbot?

Building a chatbot from scratch comes in three major phases - traditionally, each phase is a big chunk of work:

1: Natural Language Processing (NLP)

What is Natural Language Processing (NLP)?

NLP is how a computer to understand human conversations. Instead of just recognizing a few words, the computer needs to understand countless ways people might ask for the same thing.

For example, "What's the weather today?", "Is it going to rain?", and "Should I bring an umbrella?" might all be asking for the same information, but phrased differently.

Tasks under the NLP phase:

Train computers to understand human language (both text and speech)
Create thousands of example phrases to teach the computer
Extract important details from what users say (like dates, names, amounts)
Build logic that pinpoint exactly what users are trying to do
Make the system work with different languages and accents

2: Dialog Management

What is dialog management?

Dialog management is about keeping track of the entire conversation.

A waiter at a restaurant doesn't just need to understand your order; they need to remember what you've already ordered, ask follow-up questions ("How would you like that cooked?"), and handle situations when they don't understand you ("I'm sorry, could you repeat that?"). Dialog management helps chatbots do all of this!

Tasks under the dialog management phase:

Create systems that remember where you are in a conversation
Build memory so the chatbot knows what you talked about earlier
Design logic for collecting all necessary information from users
Create backup plans for when the chatbot doesn't understand you
Map out different conversation flows for various scenarios

3: Integration

What is Integration?

Once you've spent the precious time building, this phase is all about making your chatbot available to users.

Your chatbot needs to connect to real data, scale to handle many users, and have security measures in place before it's ready for the real world.

Tasks under the integration phase:

Connect the chatbot to other systems like databases or payment processors
Build security measures to protect user information
Set up infrastructure that can handle many users at once
Create tools to track how well the chatbot is performing

This entire process traditionally needs specialized experts in machine learning, linguistics, and conversational design, plus months of development time and lots of computing resources. This can be very expensive and complicated - yikes! 😱

Enter Amazon Lex... 🥁

Amazon Lex is AWS's service for building chatbots - it eliminates most of the complexity and turns chatbot building from taking months to just hours.

In Amazon Lex, building a chatbot works like this:

1: Natural Language Processing

😎 Your tasks (we do this in every project of the series):

Give Lex 10-20 example phrases of what users might say
Define the types of information your bot needs to collect

🤖 Tasks handled by Amazon Lex:

Providing pre-trained models that understand user text/speech
Generating thousands of training examples internally
Extract important details from what users say (like dates, names, amounts)
Build logic that pinpoint exactly what users are trying to do

What does this mean for me?

Instead of needing deep machine learning knowledge to build a chatbot that understands users, you just need to provide a few examples of what users might say.

2: Dialog Management

😎 Your tasks:

Define what information to continue remembering about users (we do in project 4 of the series)
Create prompts for gathering that information (we do in project 4 of the series)
Plan how the chatbot should respond if there are misunderstandings (we do in project 1 of the series)
Design the overall conversation flow (we do this in every project of this series)

🤖 Tasks handled by Amazon Lex:

Handling the conversation flow i.e. making sure the chatbot actually says 'x' after 'y'
Tracking memory with built-in features
Running the fallback conversation flow for misunderstandings

What does this mean for me?

You don't need to write complex code to handle conversations and remember information. Instead, you just need to decide what information your chatbot needs (like account type or date of birth) and what questions to ask. Lex handles all the back-and-forth conversation and memory logic automatically!

3: Integration

😎 Your tasks:

Connect your bot to your backend systems if needed (we do this in project 3 of this series)
Test and deploy your chatbot (we do this in every project of the series)

🤖 Tasks handled by Amazon Lex:

Providing built-in AWS security
Managing the infrastructure automatically
Handling automatic scaling for many users

What does this mean for me?

You don't need to worry about servers, scaling, or complex infrastructure. Lex handles all of that for you. You can focus on making your chatbot useful and relevant to your users, and let AWS take care of making sure it runs smoothly!

Extra for Experts: How does this compare to other chatbot platforms?

While many platforms offer some level of simplification, Amazon Lex has an especially tight integration with other AWS services like Lambda for backend logic and CloudWatch for monitoring. This makes it even more powerful for organizations already using AWS. In fact, we'll get to integrate Lex with Lambda in project three of this series!

Ready for the first challenge?

You've got the theory down, now comes the fun part - building your very own AI chatbot!

Are you ready to:

🤖 Build a fully functional banking chatbot?
💬 Create natural conversational flows that feel human?
🧠 Connect your chatbot to backend systems for real functionality?
💪 Add an impressive AI project to your portfolio?

Share Your Progress!

This challenge can help you build your AWS skills, your portfolio, and your personal brand! To make the most of it, share your progress on LinkedIn or GitHub.

It's so easy to share your documentation - all you have to do is:

Select Mission Accomplished at the end of this page 👇

Download your documentation as a PDF.

Select the LinkedIn icon to open a pre-populated post, all ready to go!

Select the X at the bottom of the post - it's blocking you from uploading documentation.

Click on the plus icon at the bottom of the panel.

Then select the page icon, which helps you Add a document.

Voilà! Upload your document and give it a nice title like I'm challenging myself to build an AI chatbot!
Select Done and Post to share your progress!

p.s. Does it say "Still tasks to complete!" at the bottom of the screen?

This means you still have screenshots left to upload, or questions left to answer!

Press Ctrl+F (Windows) or Command+F (Mac) on your keyboard.
Search for the text Return to later.
Jump straight to your incomplete tasks!
🙋‍♀️ Still stuck? Ask the community!