Integrate Amazon Bedrock with Einstein
Learning Objectives
After completing this unit, you’ll be able to:
- Describe how Amazon Bedrock integrates with the Einstein Trust Layer.
- Explain real-world applications and use cases.
This module was produced in collaboration with AWS, which owns, supports, and maintains the AWS products, services, and features described here. Use of AWS products, services, and features is governed by privacy policies and service agreements maintained by AWS. Learn more about partner content on Trailhead.
Before You Start
Before you start this module, make sure you complete Amazon Bedrock Foundation Models. The work you do here builds on the concepts and work you do in that badge.
Bring Hosted Models to Einstein
The integration between Amazon Bedrock and Einstein brings models hosted on Amazon Bedrock to Salesforce. This lets you use your customer relationship management (CRM) data with models provided by leading AI companies or your own custom-tuned model.
Think of a manufacturing company interested in experimenting with a foundation model from Amazon Bedrock. The company can fine-tune the model with specific product knowledge and industry expertise, and then securely integrate it with its Salesforce environment. This means the company’s sales reps can access well-trained, finely tuned AI-powered insights and generate content while working in Sales Cloud. This combination helps reps create more effective sales proposals, better understand customer needs, and improve overall productivity.
Keep Data Secure
The integration between Amazon Bedrock and Einstein establishes a secure channel through which data and AI requests can flow. The following graphic illustrates how Amazon Bedrock integrates with Einstein.
At a highlevel, if a user initiates an AI-powered task within Salesforce—whether it’s generating content, analyzing customer data, or creating predictions—the request first passes through the Einstein Trust Layer (1). The Trust Layer validates the request, applies necessary security protocols, and ensures compliance with data privacy requirements. The request then reaches Amazon Bedrock (2), where the model processes it. The response follows the same path back through the Trust Layer.
Einstein Model Builder is the central point for managing AI models within Salesforce. It provides a low-code environment where you can register and manage your AI models. To use a model, you first register it, providing the necessary credentials and configuration details. Model Builder then connects to your registered model.
After the model is registered, it becomes available for use with Prompt Builder, a tool for building, testing, and fine-tuning AI prompts that can be used across the Salesforce Platform. With Prompt Builder, you can create a variety of generative applications for your Salesforce users and ground responses on your CRM data if needed. For more information, see The Ultimate Guide to Prompt Builder.
Dive Into Real-World Applications and Use Cases
Imagine you’re a customer service manager at an IT company, and want to integrate Amazon Bedrock with Einstein. Here are the steps you can take.
- Start by selecting or customizing an AI model in Amazon's Bedrock.
- Using the Einstein Model Builder, register your model with Salesforce.
- Through Prompt Builder, you can create specific templates for common customer scenarios such as troubleshooting issues, product upgrades, and escalation decisions.
- Your agents and colleagues can now access these AI capabilities directly in their Salesforce console. For example, when a customer emails about a technical issue, the AI can immediately suggest relevant solutions.
Let’s see how different teams can use the integration between Amazon Bedrock and Einstein.
Sales Operations
-
Prioritize leads: Analyze sales data to create predictive lead scores based on factors such as likelihood to convert, recent engagement, and more.
-
Forecast sales: Use a custom model fine-tuned on historical sales and industry data to generate accurate forecasts, detecting potential market trends or seasonal patterns.
-
Automate reporting: Automate report generation and summarize performance metrics for regular sales reviews.
Customer Service
-
Real-time support suggestion: Analyze customer queries and suggest relevant responses or troubleshooting steps.
-
Content and sentiment analysis: Route customer emails to the right team automatically.
-
Knowledge base enhancement: Refine and update the knowledge base by summarizing recent customer interactions, automating article creation, and filling gaps based on customer issues.
Marketing
-
Campaign content generation: Generate campaign materials such as emails, social media posts, and ad copy that align with messaging on what resonates with each customer group.
-
Predictive insights: Predict which marketing campaign will work best based on CRM and historical data.
-
Automated tasks: Automate repetitive marketing tasks such as project management, SEO audits, and more.
Wrap Up
With Amazon Bedrock and Einstein, you can bring powerful custom AI models into your Salesforce environment. This integration lets you enhance your CRM with base or customized AI models from Amazon and other leading providers such as Anthropic, Cohere, AI21 Labs, and more. In the next unit, you dive further into integration.