Connect the SEA Explorer demo app

Now that you have a working agent with Signals attributes and AgentCore Memory, you can connect it to a frontend application. The SEA Explorer demo app is a Next.js travel site that calls Bedrock directly using the same model and credentials from the notebook.

The app includes a chat interface where users can ask travel questions. As users browse destinations and experiences, Snowplow tracking generates the behavioral events that Signals processes into the attributes your agent uses for personalization. AgentCore Memory persists context across chat sessions.

Prerequisites

Before connecting the demo app, ensure you have:

• Completed the previous steps in this accelerator (agent, Signals attributes, and memory)
• Your AWS credentials - AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY - with Bedrock access
• Your AWS region - the same one used in the notebook
• Your AgentCore Memory ID - printed when you created the memory resource
• Your Signals credentials - the same values you configured in the notebook's credentials cell

Run the demo app

Create a file called .env with your credentials from the notebook. Include your AWS credentials, memory ID, and Signals connection:

CDI

bash

AI_MODEL_PROVIDER=bedrock
AI_MODEL_NAME=us.anthropic.claude-sonnet-4-20250514-v1:0
AWS_REGION=us-west-2
AWS_ACCESS_KEY_ID=your_access_key
AWS_SECRET_ACCESS_KEY=your_secret_key
AGENTCORE_MEMORY_ID=your_memory_id

NEXT_PUBLIC_SNOWPLOW_SIGNALS_API_URL=https://example.signals.snowplowanalytics.com
SNOWPLOW_SIGNALS_API_KEY=your_api_key
SNOWPLOW_SIGNALS_API_KEY_ID=your_api_key_id
SNOWPLOW_SIGNALS_ORGANIZATION_ID=your_org_id
SNOWPLOW_SIGNALS_SERVICE_NAME=travel_service

Sandbox

bash

AI_MODEL_PROVIDER=bedrock
AI_MODEL_NAME=us.anthropic.claude-sonnet-4-20250514-v1:0
AWS_REGION=us-west-2
AWS_ACCESS_KEY_ID=your_access_key
AWS_SECRET_ACCESS_KEY=your_secret_key
AGENTCORE_MEMORY_ID=your_memory_id

NEXT_PUBLIC_SNOWPLOW_SIGNALS_API_URL=https://you.signals.snowplowanalytics.com
SNOWPLOW_SIGNALS_TRIAL_TOKEN=your_token
SNOWPLOW_SIGNALS_SERVICE_NAME=travel_service

Pull and run the published Docker image, passing your credentials via the env file:

bash

docker run --env-file .env -p 3000:3000 snowplow/demo-travel-site:accelerator

Open http://localhost:3000 in your browser.

Verify configuration

The app header includes two read-only status dropdowns:

• Signals — shows whether Signals is connected in CDI or Sandbox mode, and which credentials are configured
• AgentCore — shows the current Bedrock model, AWS region, and memory ID

If either dropdown shows "Not set" for required fields, check that the corresponding environment variables are present in your .env file and restart the container.

Test the integration

1. Chat with the agent: click the chat icon in the bottom right corner and ask a travel question. The agent will call Bedrock directly with the same model used in the notebook (us.anthropic.claude-sonnet-4-20250514-v1:0).

2. Test memory: ask the agent for destination recommendations. Because you seeded customer history in the notebook, the agent should reference preferences for warm, adventurous, family-friendly destinations without you needing to state them.

3. Test Signals: browse some destination and experience pages to generate tracking events. Then ask the agent "Based on my browsing behavior, what experiences would suit me?" The agent will fetch your behavioral profile from Signals and personalize its response.

tip

All configuration is managed through environment variables. To change settings, update your .env file and restart the container.