From "Broadcast" to "Individual"

In the old world of business, you sent the same catalog to 1 million people. In the AWS AI world, you give 1 million people their own "Store Front" tailored to their specific tastes.

To do this, AWS uses two very different techniques: Machine Learning Recommendations and Generative AI Content.

---

1. Recommendation Engines: Amazon Personalize

Amazon Personalize is a fully managed service that allows you to build the same recommendation technology used by Amazon.com.

How it works:

Unlike a simple "People who bought x also bought y" rule, Personalize looks at the Sequences of events. It understands that:

• User A looks at hiking boots, then socks, then a compass.
• User B looks at hiking boots, then a dress, then high heels.
• Even though they both looked at boots, User A is a "Hiker" and User B is a "Fashionista." Personalize adjusts their next recommendation accordingly.

Use Case: A movie streaming app (like Netflix) showing you "Movies You'll Love" based on your specific watching history.

---

2. Generative Content: Amazon Bedrock

While Personalize picks which product to show, Amazon Bedrock can create the message for that product.

The "Dynamic Ad" Use Case:

Imagine an e-commerce site:

1. Personalize identifies that a customer loves "Environmentally Friendly" products.
2. Bedrock triggers an LLM to rewrite a generic product description into one that focuses on "Sustainability."
3. Bedrock (Titan Image) generates a background for the product image showing it in a lush, green forest.

The Result: The customer sees an ad that feels like it was written specifically for them (because it was).

---

3. High-Level Comparison

Feature	Amazon Personalize	Amazon Bedrock (GenAI)
Goal	Selection (What should I show?)	Creation (How should I say it?)
Logic	Ranking and Sorting data	Probability and Creation
Data Needed	User clickstreams / history	Contextual prompts

graph LR
    A[User Profile] --> B[Amazon Personalize]
    B -->|Selection| C[Product: Eco-Friendly Jacket]
    C --> D[Amazon Bedrock]
    D -->|Generation| E[Custom Email: 'The greenest jacket for your next hike!']
    E --> F[Happy Customer]

---

4. Summary: Maximizing Customer Relevance

The value of AI in marketing is Relevance.

• Irrelevant ads are "Spam."
• Relevant ads are "Service." By using Personalize and Bedrock, companies can move from being a "Spammer" to being a "Digital Concierge."

---

Exercise: Identify the Service

A music streaming startup wants to create a "Discover Weekly" playlist for their users. They have billions of data points showing what users listen to and what they skip. They don't have a team of machine learning researchers. Which service should they use to build the recommendation engine?

• A. Amazon SageMaker.
• B. Amazon Personalize.
• C. Amazon Rekognition.
• D. Amazon Lex.

The Answer is B! Amazon Personalize is the "Managed" recommendation service that allows developers to build high-quality suggestions without needing a deep ML background.

---

Knowledge Check

What's Next?

Content is exciting, but what about the "Boring" data trapped in messy documents? In the next lesson, we see how to unlock the "Company Brain." Find out in Lesson 3: AI for Document Processing and Search.