How to Launch Your Own Private ChatGPT Interface with AWS Bedrock

Raymond Yeh
Wielded
Published in
5 min readJan 26, 2024

--

ChatGPT has taken the world by storm, showcasing the immense potential of large language models. However, the free research version has significant limitations — slow response times, limited daily usage, lack of context, and no customization.

What if you could launch your own private ChatGPT interface tailored to your team’s needs? With AWS Bedrock, now you can!

In this comprehensive guide, I’ll walk you through exactly how to set up Claude, Anthropic’s powerful ChatGPT alternative, on AWS Bedrock and integrate it with Wielded to unlock enterprise productivity.

About Claude & Wielded

The Power Duo: Claude + Wielded is one of the most advanced setup for generative AI assistants today.

Developed by AI safety startup Anthropic using their cutting-edge Constitutional AI technique, Claude overcomes many of the issues with unreliable and potentially harmful AI systems.

Some key advantages that make Claude stand out:

  • Massive 200,000 token context window (2X of GPT-4) to deeply understand documents and conversations without losing track of context.
  • Built-in safety constraints through self-supervision during training to minimize harmful, biased and incorrect outputs.
  • Advanced techniques like information retrieval augmentation to incorporate external knowledge sources accurately.
  • Easy customization through fine-tuning and examples to adapt Claude’s behavior as per user preferences.

Paired with Wielded’s ChatGPT interface and innovative features like custom personas and team billing, you get the best of both worlds — a secure, customized and context-aware AI assistant tailored to your organization’s needs:

  • Blazing fast response times with Claude hosted on AWS global infrastructure
  • Huge context window to deeply understand projects and documents
  • Custom personalities to switch contexts while retaining preferences
  • Team access with centralized billing to simplify procurement
  • Privacy as all data stays within your AWS account per region

Getting Access to Claude on AWS Bedrock

Getting Started with AWS Bedrock AWS Bedrock makes cutting-edge generative AI models easily accessible through a unified API. Through a simple sign-up process, you can deploy private instances of models like Claude to augment your team’s workflow.

Here are the key steps:

1. Sign up for an AWS account

As a prerequisite, you need an AWS account with programmatic access enabled. If you don’t have one already, simply go to aws.amazon.com to create a free account.

2. Request access for Claude in the AWS Bedrock console

Request for Claude mode, you may need to provide a use case

Once signed in, use the search bar to navigate to the Bedrock console. Click on “Request model access” and submit your use case for Claude. Do note Claude is currently only available in the US East and US West regions.

3. Get approved (typically 24–48 hours)

After submitting the request, the AWS team will manually review and approve access usually within a day. You’ll receive an email from AWS when access is granted.

4. Create an IAM user to generate API keys

Attach only the AmazonBedrockFullAccess policy to the user

Under AWS Identity and Access Management (IAM), create a new programmatic user exclusively for Bedrock access. Choose the “AmazonBedrockFullAccess” policy and save the access and secret key in a secure place.

5. Save keys in a secure place for integration

Make sure to copy the access and secret key — it’s impossible to retrieve the secret key later. You’ll need both for integrating Claude with applications like Wielded.

Video Walkthrough

Check out this step-by-step video walkthrough for more detailed instructions on gaining access and using the Bedrock playground.

Integrating Claude with Wielded

Once you have API access keys for Claude, it’s simple to connect it with Wielded in just a few clicks:

1. Create a Wielded account

First, sign up for a free Wielded account via the website and confirm your email address.

2. Navigate to Integrations and select AWS Bedrock

In Account Settings, choose the Integrations tab and click on New Integration. Select “AWS Bedrock” as the provider.

3. Enter your AWS keys and select Claude

Copy the access and secret keys from the IAM user created earlier. Next, choose Claude v2.1 as the foundation model and enter the AWS region.

4. Click “Add AWS Integration”

Double check the keys and model information entered are correct before hitting save to complete the integration process.

And you’re all set! The Wielded chat interface will now privately query your dedicated Claude instance to supercharge productivity.

You can repeat the process to add other models like Claude Instant as well for testing different versions. With the integration complete, your team can now leverage Claude’s advanced capabilities through Wielded’s innovative features.

Unleash Your Team’s Potential Generative AI is fundamentally changing how we work. With Claude and Wielded, you now have access to:

  • A context-aware assistant that deeply understands your team’s projects for accurate and relevant outputs
  • Custom personalities tailored to individual preferences while retaining organizational knowledge
  • Secure private deployment without relying on external services
  • Centralized team access with consolidated billing and usage analytics

Here are just some of the ways Claude on AWS Bedrock can be transformative:

Marketing & Content Creation

  • Craft high-converting marketing copy tailored to your offering
  • Conduct market analysis to optimize positioning
  • Ideate and outline blog posts around trending topics
  • Rewrite content for search engine optimization
  • Automate social media post creation based on performance

Customer Support

  • Quickly research to provide in-depth answers for customer queries
  • Analyze support tickets to identify recurring issues
  • Generate article ideas to address frequent questions
  • Create canned responses for common requests

Legal Services

  • Review and summarize terms across contracts
  • Flag potential issues requiring closer inspection
  • Accelerate research on case law and precedents
  • Generate first drafts for legal documents

Recruiting & HR

  • Source qualified candidates proactively based on role needs
  • Review resumes and profiles to schedule interviews
  • Analyze job post performance and suggest optimizations
  • Craft engaging job descriptions tailored to the position

The possibilities are endless — marketing, support, legal, recruiting, finance, operations, product — every team can benefit from having Claude on tap!

So don’t get left behind in the AI productivity race! Set up Claude on AWS Bedrock today and elevate team performance to the next level with Wielded.

--

--

Raymond Yeh
Wielded

Founder @ wielded.com - Generative AI Workspace for Modern Teams