Talk2Data - Data Intelligence AI Agent

DEPLOYMENT

Container port: 8080 Health check endpoint: GET /ping Architecture: ARM64

Prerequisites:

1. Ensure Qwen3 32B is available in Amazon Bedrock for your account/region (check Bedrock Console > Model access; request access if not already enabled).
2. Attach an IAM role to your AgentCore Runtime execution role with the following permission: { "Effect": "Allow", "Action": "bedrock:InvokeModel", "Resource": "arn:aws:bedrock:*::foundation-model/qwen.qwen3-32b-v1:0" }
3. Deploy this container image to Amazon Bedrock AgentCore Runtime.

Environment variables:

• AWS_REGION (optional): AWS region for Bedrock calls. Defaults to "eu-central-1".
• BEDROCK_MODEL_ID (optional): Bedrock model identifier. Defaults to "qwen.qwen3-32b-v1:0".
• Database connection (optional): Set default DB connection via environment variables, or supply per-request (see USAGE below). Required fields if using env vars: DB_HOST, DB_PORT, DB_NAME, DB_USER, DB_PASSWORD.

Example deployment (AgentCore Runtime, via console or CLI): Point the AgentCore Runtime container image field to this product's image URI, attach the IAM role described above, and set any environment variables needed.

USAGE

1. Health check: GET /ping Response: {"status": "Healthy"}

2. Ask a natural-language question about your data: POST /invocations Body: { "prompt": "What were total sales last quarter?", "db": { "host": "your-db-host", "port": 5432, "dbname": "your-db-name", "user": "your-db-user", "password": "your-db-password" } }

Response: { "sql": "<generated SQL query>", "results": [ ... structured query results ... ] }

Note: the "db" field is optional if default database connection environment variables are configured; otherwise it is required per request.

SUPPORT

For support or questions, contact [work@mejuvante.co.in ].