Today, we're excited to announce Retool’s integration with Amazon Bedrock and Knowledge Bases for Amazon Bedrock. This integration enables businesses to leverage the power and security of Amazon Bedrock with the ease and speed of development on Retool to build AI applications.
For companies with strict data policies, sending data to a third-party LLM can be a dealbreaker. With Amazon Bedrock, businesses can host LLMs within their VPC, ensuring compliance with their security and governance requirements. Retool customers who have adopted AWS can take advantage of all that Bedrock has to offer while keeping their data safe and secure.
The Amazon Bedrock integration is currently available to Retool cloud customers and scheduled for general availability for self-hosted customers in our June 4, 2024 release.
Data security is top of mind for many companies looking to integrate AI into their applications. AI applications are more powerful when they have access to internal data, but connecting directly with third-party APIs often means sharing sensitive information outside the safety of their VPC. Many companies turn to Amazon Bedrock to ensure data remains secure within the AWS ecosystem through encryption, VPC integration, private endpoints, and access controls.
Amazon Bedrock provides fully managed access to leading AI models like Anthropic’s Claude, Meta’s Llama 3, Mistral, Cohere’s Command, and Amazon’s Titan through a single service. Bedrock’s serverless design means that users can experiment, train, and evaluate models without managing any infrastructure. Businesses can also use their existing Knowledge Bases to provide models with data that’s tailored to their organization and specific use cases, while keeping data contained within their VPC.
Bedrock is a powerful tool for building and evaluating language models—but building a full application around a model can take weeks or even months if developers have to start from scratch. This process often involves integrating with multiple APIs, managing API keys, and building custom tools and UIs around LLM responses–requiring a specialized skill set and collaboration from data, machine learning, and full-stack engineering teams to develop the application.
Retool AI helps bridge the gap by making tasks like text classification, generating AI chat responses, and document and image processing more approachable through:
- An all-in-one solution: Retool combines all the necessary tools to build AI applications in one centralized platform, streamlining your development process.
- Model switching made easy: Easily switch between different AI models by selecting the model of choice through a drop-down menu, eliminating the need for multiple API calls.
- Seamless data integration: Integrate internal data with third-party LLMs using Retool Vectors, with a single click, providing context and increasing model performance.
- Pre-built AI components and workflows: Retool's AI actions and workflow building blocks allow easy integration of AI capabilities into a myriad of applications and processes.
The Retool and Amazon Bedrock integration allows users to access their Bedrock models in Retool’s AI actions and connect a Knowledge Base through Retool Vectors. This combination enables the development of AI applications that are connected to internal corporate data and workflows with ease, speed, and security.
A key feature of this integration is the ability to customize foundational models with an Amazon Knowledge Base using Retool Vectors and the retrieval-augmented generation (RAG) technique.
Amazon has some great informative resources if you're interesting in learning more about RAG.
- Connect your Amazon Bedrock models: Start by integrating your Amazon Bedrock models with Retool. Authentication is done securely through IAM and enables access to AI models like Anthropic’s Claude, Meta’s Llama 3, Mistral, Cohere’s Command, and Amazon’s Titan.
- Connect your Amazon Knowledge Bases: Enhance out-of-the-box LLMs with data from your Amazon Knowledge Bases. Create a new vector in Retool Vectors and import an Amazon Knowledge Base. This will be used to provide context for third party LLMs through RAG.
- Build and customize your app: Use Retool's intuitive IDE and workflow automation products to create and customize your model through a Retool AI building block. Easily switch between Bedrock models and tailor the AI capabilities to your specific needs.
- Deploy securely: Retool gives business the option to deploy your application on your own infrastructure using Retool’s self-hosted feature. This option is a great fit for businesses that require their data to stay within their VPC.
![Accessing Bedrock, our newest AI integration](/blog/_next/image?url=https%3A%2F%2Fcdn.sanity.io%2Fimages%2Fbclf52sw%2Fproduction%2F6360e5663a1c2a9e7d969af9c76a8c0e77e6f29b-1920x1080.png&w=3840&q=75)
The Retool and Amazon Bedrock integration empowers businesses to securely develop and deploy AI applications with ease. By combining the strengths of both platforms, users can build applications fast and get ROI on their Bedrock models in days, not months.
To start using this integration:
- For cloud customers: You can try out the integration today. Simply enable your Amazon Bedrock models to connect to Retool through IAM on Retool AI, and connect an Amazon Knowledge Base to Retool Vectors.
- For self-hosted customers: This integration is scheduled for general availability in our June 4, 2024 release.
Reader