Create generative AI brokers that work together together with your firms’ programs in just a few clicks utilizing Amazon Bedrock in Amazon SageMaker Unified Studio

Right now we’re saying that basic availability of Amazon Bedrock in Amazon SageMaker Unified Studio.

Corporations of all sizes face mounting strain to function effectively as they handle rising volumes of information, programs, and buyer interactions. Handbook processes and fragmented info sources can create bottlenecks and gradual decision-making, limiting groups from specializing in higher-value work. Generative AI brokers provide a robust answer by routinely interfacing with firm programs, executing duties, and delivering immediate insights, serving to organizations scale operations with out scaling complexity.

Amazon Bedrock in SageMaker Unified Studio addresses these challenges by offering a unified service for constructing AI-driven options that centralize buyer information and allow pure language interactions. It integrates with current purposes and consists of key Amazon Bedrock options like basis fashions (FMs), prompts, information bases, brokers, flows, analysis, and guardrails. Customers can entry these AI capabilities by their group’s single sign-on (SSO), collaborate with workforce members, and refine AI purposes while not having AWS Management Console entry.

Generative AI-powered brokers for automated workflows

Amazon Bedrock in SageMaker Unified Studio permits you to create and deploy generative AI brokers that combine with organizational purposes, databases, and third-party programs, enabling pure language interactions throughout the complete know-how stack. The chat agent bridges complicated info programs and user-friendly communication. By utilizing Amazon Bedrock features and Amazon Bedrock Knowledge Bases, the agent can join with information sources like JIRA APIs for real-time undertaking standing monitoring, retrieve buyer info, replace undertaking duties, and handle preferences.

Gross sales and advertising groups can rapidly entry buyer info and their assembly preferences, and undertaking managers can effectively handle JIRA duties and timelines. This streamlined course of enhances productiveness and buyer interactions throughout the group.

The next diagram illustrates the generative AI agent answer workflow.

Answer overview

Amazon Bedrock offers a ruled collaborative surroundings to construct and share generative AI purposes inside SageMaker Unified Studio. Let’s have a look at an instance answer for implementing a buyer administration agent:

• An agentic chat could be constructed with Amazon Bedrock chat purposes, and built-in with features that may be rapidly constructed with different AWS companies equivalent to AWS Lambda and Amazon API Gateway.
• SageMaker Unified Studio, utilizing Amazon DataZone, offers a complete information administration answer by its built-in companies. Group directors can management member entry to Amazon Bedrock fashions and options, sustaining safe identification administration and granular entry management.

Earlier than we dive deep into the deployment of the AI agent, let’s stroll by the important thing steps of the structure, as proven within the following diagram.

The workflow is as follows:

1. The person logs into SageMaker Unified Studio utilizing their group’s SSO from AWS IAM Identity Center. Then the person interacts with the chat utility utilizing pure language.
2. The Amazon Bedrock chat utility makes use of a operate to retrieve JIRA standing and buyer info from the database by the endpoint utilizing API Gateway.
3. The chat utility authenticates with API Gateway to securely entry the endpoint with the random API key from AWS Secrets Manager, and triggers the Lambda operate based mostly on the person’s request.
4. The Lambda operate performs the actions by calling the JIRA API or database with the required parameters offered from the agent. The agent has the potential to:

1. 1. Present a short buyer overview.
   2. Listing latest buyer interactions.
   3. Retrieve the assembly preferences for a buyer.
   4. Retrieve open JIRA tickets for a undertaking.
   5. Replace the due date for a JIRA ticket.

Conditions

You want the next conditions to observe together with this answer implementation:

We assume you’re accustomed to elementary serverless constructs on AWS, equivalent to API Gateway, Lambda features, and IAM Id Middle. We don’t deal with defining these companies on this put up, however we do use them to point out use circumstances for the brand new Amazon Bedrock options inside SageMaker Unified Studio.

Deploy the answer

Full the next deployment steps:

1. Obtain the code from GitHub.
2. Get the worth of JIRA_API_KEY_ARN, JIRA_URL, and JIRA_USER_NAME for the Lambda operate.
3. Use the next AWS CloudFormation template, and seek advice from Create a stack from the CloudFormation console to launch the stack in your most popular AWS Area.
4. After the stack is deployed, observe down the API Gateway URL worth from the CloudFormation Outputs tab (ApiInvokeURL).
5. On the Secrets and techniques Supervisor console, discover the secrets and techniques for JIRA_API_KEY_ARN, JIRA_URL, and JIRA_USER_NAME.
6. Select Retrieve secret and duplicate the variables from Step 2 to the key plaintext string.
7. Register to SageMaker Unified Studio utilizing your group’s SSO.

Create a brand new undertaking

Full the next steps to create a brand new undertaking:

1. On the SageMaker Unified Studio touchdown web page, create a new project.
2. Give the undertaking a reputation (for instance, crm-agent).
3. Select Generative AI utility improvement profile and proceed.
4. Use the default settings and proceed.
5. Assessment and select Create undertaking to substantiate.

Construct the chat agent utility

Full the next steps to build the chat agent application:

1. Below the New part situated to the correct of the crm-agent undertaking touchdown web page, select Chat agent.

It has a listing of configurations on your agent utility.

2. Below the mannequin part, select a desired FM supported by Amazon Bedrock. For this crm-agent, we select Amazon Nova Professional.
3. Within the system immediate part, add the next immediate. Optionally, you can add examples of person enter and mannequin responses to enhance it.

You're a buyer relationship administration agent tasked with serving to a gross sales particular person plan their work with prospects. You're supplied with an API endpoint. This endpoint can present info like firm overview, firm interplay historical past (assembly occasions and notes), firm assembly preferences (assembly sort, day of week, and time of day). You can too question Jira duties and replace their timeline. After receiving a response, clear it up right into a readable format. If the output is a numbered listing, format it as such with newline characters and numbers.

4. Within the Features part, select Create a brand new operate.
5. Give the operate a reputation, equivalent to crm_agent_calling.
6. For Perform schema, use the OpenAPI definition from the GitHub repo.

7. For Authentication technique, select API Keys (Max. 2 Keys)and enter the next particulars:
   1. For Key despatched in, select Header.
   2. For Key identify, enter x-api-key.
   3. For Key worth, enter the Secrets and techniques Supervisor api Key
8. Within the API servers part, enter the endpoint URL.
9. Select Create to complete the operate creation.
10. Within the Features part of the chat agent utility, select the operate you created and select Save to complete the appliance creation.

Instance interactions

On this part, we discover two instance interactions.

Use case 1: CRM analyst can retrieve buyer particulars saved within the database with pure language.

For this use case, we ask the next questions within the chat utility:

• Give me a short overview of buyer C-jkl101112.
• Listing the final 2 latest interactions for buyer C-def456.
• What communication technique does buyer C-mno131415 favor?
• Suggest optimum time and phone channel to achieve out to C-ghi789 based mostly on their preferences and our final interplay.

The response from the chat utility is proven within the following screenshot. The agent efficiently retrieves the client’s info from the database. It understands the person’s query and queries the database to seek out corresponding solutions.

Use case 2: Undertaking managers can listing and replace the JIRA ticket.

On this use case, we ask the next questions:

• What are the open JIRA Duties for undertaking id CRM?
• Please replace JIRA Activity CRM-3 to 1 weeks out.

The response from the chat utility is proven within the following screenshot. Much like the earlier use case, the agent accesses the JIRA board and fetches the JIRA undertaking info. It offers a listing of open JIRA duties and updates the timeline of the duty following the person’s request.

Clear up

To keep away from incurring extra prices, full the next steps:

1. Delete the CloudFormation stack.
2. Delete the operate element in Amazon Bedrock.
3. Delete the chat agent utility in Amazon Bedrock.
4. Delete the domains in SageMaker Unified Studio.

Value

Amazon Bedrock in SageMaker Unified Studio doesn’t incur separate costs, however you can be charged for the person AWS companies and assets utilized inside the service. You solely pay for the Amazon Bedrock assets you utilize, with out minimal charges or upfront commitments.

When you want additional help with pricing calculations or have questions on optimizing prices on your particular use case, please attain out to AWS Help or seek the advice of together with your account supervisor.

Conclusion

On this put up, we demonstrated find out how to use Amazon Bedrock in SageMaker Unified Studio to construct a generative AI utility to combine with an current endpoint and database.

The generative AI options of Amazon Bedrock remodel how organizations construct and deploy AI options by enabling speedy agent prototyping and deployment. Groups can swiftly create, take a look at, and launch chat agent purposes, accelerating the implementation of AI options that automate complicated duties and improve decision-making capabilities. The answer’s scalability and adaptability enable organizations to seamlessly combine superior AI capabilities into current purposes, databases, and third-party programs.

By a unified chat interface, brokers can deal with undertaking administration, information retrieval, and workflow automation—considerably lowering guide effort whereas enhancing person expertise. By making superior AI capabilities extra accessible and user-friendly, Amazon Bedrock in SageMaker Unified Studio empowers organizations to realize new ranges of productiveness and buyer satisfaction in right this moment’s aggressive panorama.

Check out Amazon Bedrock in SageMaker Unified Studio on your personal use case, and share your questions within the feedback.

In regards to the Authors

Jady Liu is a Senior AI/ML Options Architect on the AWS GenAI Labs workforce based mostly in Los Angeles, CA. With over a decade of expertise within the know-how sector, she has labored throughout numerous applied sciences and held a number of roles. Captivated with generative AI, she collaborates with main shoppers throughout industries to realize their enterprise targets by growing scalable, resilient, and cost-effective generative AI options on AWS. Exterior of labor, she enjoys touring to discover wineries and distilleries.

Justin Ossai is a GenAI Labs Specialist Options Architect based mostly in Dallas, TX. He’s a extremely passionate IT skilled with over 15 years of know-how expertise. He has designed and applied options with on-premises and cloud-based infrastructure for small and enterprise firms.