Integration with AWS Bedrock
Last updated
Last updated
AWS Bedrock provides a platform for building and deploying AI models. By integrating AWS Bedrock Agents into MangoApps AI Studio as Assistants, you can enhance automation and intelligence within your intranet environment.
If your company already has a configured and licensed AI Assistant from AWS Bedrock, you can integrate it into your MangoApps domain.
An active AWS account with access to Amazon Bedrock.
Administrator access to MangoApps AI Studio.
AWS IAM permissions to create and manage Bedrock Agents.
Log into AWS Management Console and navigate to Amazon Bedrock.
Click on Agents and select the Create Agent button if you do not already have an active Agent.
Continue through the steps to create an agent:
Choose a Foundation Model from the Model Catalog.
Configure your Agent with Action Groups and define its response flows.
Deploy the Agent by associating an Alias with a specific version.
Test the Agent in real-time before deployment.
From the search bar, navigate to the AWS IAM.
From the IAM dashboard, click the My security credentials from the right hand Quick Links menu.
Create an Access Key with policies granting access to Amazon Bedrock services. Attaching all the necessary Bedrock-related permissions.
Generate an Access Key and Secret Key. These will be used on the MangoApps side.
Once added, you will be able to select Amazon Bedrock Agents from the AI Service Provider dropdown menu.
Navigate to the AI Service Providers tab within the AI Studio module. Click the Connect an AI Service Provider button in the upper right hand corner.
Select AWS Bedrock from the AI Service dropdown menu, then create a Name and Icon. This name and icon will appear in the AI Service Providers list. Once AWS Bedrock is added to the list, it will no longer be available in the dropdown menu. To add a new AWS Bedrock instance, you must first remove the existing one within MangoApps.
Enter in the AWS key and Secret token generated in the AWS setup steps above. For security purposes, the key and token will not be displayed in MangoApps.
Click the Create an Assistant button and choose Connect Your Assistant from the dropdown menu. This will take you to the assistant configuration screen.
The Configure Assistant menu will be very similar to the Create an Assistant menu. Complete the required fields and confirm that your provider has been successfully listed under the AI Service Provider tab.
Within the Customize Assistant section, select AWS Bedrock from the AI Service Provider dropdown. Once selected, choose the AWS Bedrock Agent you would like to connect as a MangoApps AI Assistant from the dropdown menu.
Finally, complete the remaining menu configurations. Since the AI service provider supplies the LLM, no additional configuration is required in this menu.
Your Assistant is now connected. By default, it will remain disabled until enabled from the Admin Menu.
Access Control: Use IAM roles and policies to restrict access to AWS Bedrock agents.
Logging & Monitoring: Enable AWS CloudTrail and MangoApps Audit Logs to track AI interactions.
Compliance: Validate AI outputs against organizational policies to prevent misinformation.
Pilot Testing: Start with a small user group to validate AI responses and improve accuracy.
User Training: Educate employees on AI functionalities and best practices.
Feedback Loop: Continuously refine the AI model based on user feedback.
Gradual Expansion: Scale up deployment based on adoption rates and performance analysis.
To disconnect the AWS Bedrock LLM from MangoApps, navigate to Modules > AI Studio > AI Service Providers within the Admin Portal. Click the 3-Dot action menu next to AWS Bedrock and click Remove LLM from the action options.
