Why should you care?

AwsVerse support chatbot summarizes tickets and drafts replies using Claude on Bedrock with enterprise data boundaries.

See it live — copy this example

Run in AWS CloudShell / local AWS CLI v2, or follow the matching steps in the AWS Console (Free Tier).

import boto3
bedrock = boto3.client('bedrock-runtime', region_name='ap-south-1')
resp = bedrock.invoke_model(
    modelId='anthropic.claude-3-haiku-20240307-v1:0',
    body='{"anthropic_version":"bedrock-2023-05-31","max_tokens":256,"messages":[{"role":"user","content":"Summarize AwsVerse refund policy in 3 bullets."}]}'
)
print(resp['body'].read())

What happened?

• boto3 calls Bedrock runtime with Claude Haiku.
• IAM policy bedrock:InvokeModel scoped to approved model IDs only.

Practice next

1. Request model access in Bedrock console.
2. Create IAM policy for invoke on one model.
3. Run Python script from EC2 or Lambda.
4. Add guardrails blocking sensitive topics.
5. Compare Haiku vs Sonnet latency and cost.

Remember

Bedrock = managed foundation models. IAM controls model access. Monitor invocation metrics.

AwsVerse support assist

Agents spend 10 min reading ticket history.

Outcome: Bedrock summary cuts handle time 30% with human review.