AWS Builder Center
Weekend Creative Agent Challenge: DailyMuse AI

Weekend Creative Agent Challenge: DailyMuse AI

DailyMuse AI is an always-on creative agent built for the AWS Weekend Creative Agent Challenge. It creates personalized poems, stories, ideas, and other creative content automatically using generative AI and AWS serverless services. The agent works in the background, so a fresh creation is ready when you return. This article covers the vision, development process, AWS architecture, challenges, and key learnings from building the project.

Hello Everyone I am Ritik Dhanotiya AWS Student Builder Group Leader | Ex- AWS SBCL

Vision & What It Does

The idea behind DailyMuse AI is simple: creativity should not always require us to sit down, open an application, and start from scratch. Sometimes the best creative experience is the one that is already waiting for you.
For the Weekend Creative Agent Challenge, I built DailyMuse AI, an always-on creative agent designed to generate personalized creative content automatically. Instead of being just another AI text generator that waits for a user prompt, DailyMuse is designed around the idea of an autonomous creative assistant that can prepare something new for the user and have it ready when they return.
DailyMuse allows users to configure their creative preferences, including the type of content they enjoy, creative styles, themes, and additional instructions. Based on these preferences, the agent can create different types of content such as poems, short stories, quotes, creative ideas, and other creative concepts.
The key difference is the always-on agent workflow. The user does not need to keep the application open for the agent to work. A scheduled AWS workflow can trigger the agent, retrieve the user's preferences, build a contextual prompt, generate new content using an AI model, and store the resulting creation. When the user returns to DailyMuse, the creation is already available on the dashboard.
The application also includes features such as a creative history, manual AI generation, remixing existing creations, feedback, creative preferences, and agent controls. I intentionally designed the application around real data rather than fake dashboard statistics. For example, streaks, total creations, agent status, and history are based on actual application data rather than predefined demo values.



How I Built It

I started by defining the core idea around the challenge requirement: "The best tool is the one you never have to open." This immediately pushed the project beyond a traditional AI generation interface. Instead of focusing only on a prompt box, I designed the application around an autonomous workflow.
The frontend was built as a modern responsive web application with a clean AI-product interface. I created separate sections for the dashboard, creation workflow, history, discover/community functionality, and settings. The dashboard focuses on the user's current creative state and shows the latest creation, agent status, statistics, and recent activity.
One of the important design decisions was removing the traditional sign-in and sign-up flow for the MVP. I wanted users to be able to open the application and start using it immediately. Instead, the application uses a persistent anonymous user ID stored locally in the browser. This ID allows preferences and creations to be associated with the same user without adding unnecessary authentication complexity to the challenge MVP.
Another important decision was to avoid fake data. During development, I made sure that dashboard information such as total creations, streaks, favorite styles, history, and agent execution status would represent actual application data. When there is no data, the application displays an appropriate empty state instead of inventing numbers.
One of the biggest challenges was making the application genuinely functional rather than just creating a visually impressive frontend. I had to trace the complete flow from the React interface to the backend API, AI generation service, database, and finally back to the frontend. This required paying close attention to API configuration, request handling, response parsing, error states, and persistence.
Another challenge was deployment. Since the application is a React-based single-page application, direct navigation to routes such as  /create ,  /history , and  /settings  initially resulted in a Netlify 404. I fixed this by configuring Netlify's SPA redirect so that application routes are handled correctly by React Router.
The most important architectural decision was keeping the autonomous agent independent from the browser. I did not want the agent to depend on a browser timer or the user keeping the website open. Instead, the scheduled agent workflow is designed to run on AWS infrastructure.



AWS Services Used / Architecture Overview

The project uses AWS services to support the backend and autonomous agent workflow.

Amazon Bedrock

Amazon Bedrock provides the AI generation capability. It is responsible for generating the creative content based on the user's preferences, context, and instructions.

AWS Lambda

AWS Lambda handles the serverless backend logic. It processes API requests, communicates with the database and AI service, and runs the autonomous daily generation workflow.

Amazon DynamoDB

DynamoDB stores the application's real data, including:
  • Anonymous user profiles
  • Creative preferences
  • Generated creations
  • Feedback
  • Agent execution information
  • Community prompts
This allows the application to persist user activity and retrieve it whenever the user returns.

Amazon EventBridge

Amazon EventBridge Scheduler is used to trigger the autonomous creative agent. This is an important part of the project because the generation process does not depend on the user's browser being open.
The high-level architecture is:
User
│
▼
Web Application
│
▼
Backend API
│
┌─────────┴─────────┐
▼ ▼
AWS Lambda DynamoDB
│ ▲
▼ │
Amazon Bedrock │
│ │
└─────────┬─────────┘
│
▼
Creative Output

Autonomous Agent Workflow

EventBridge Scheduler
│
▼
Daily Generation
Lambda
│
▼
DynamoDB
│
▼
Prompt Builder
│
▼
Amazon Bedrock
│
▼
Validation
│
▼
DynamoDB
│
▼
Creation Ready
The scheduled workflow checks active agents, retrieves their preferences and relevant previous information, generates a new creation, validates the result, and stores it. Duplicate daily generations are prevented by checking whether a creation already exists for that user and date.
The frontend is deployed as a web application, while the autonomous workload is handled by AWS serverless infrastructure. This separation means the agent can continue working independently of the user's browser session.



What I Learned

This challenge taught me that building an AI agent is very different from simply connecting an application to an AI API.
The biggest lesson was understanding the importance of workflow design. A useful agent needs context, persistence, scheduling, error handling, and a clear purpose. The AI model itself is only one part of the system.
I also learned more about designing serverless architectures with AWS. Working with Lambda, DynamoDB, EventBridge, and Amazon Bedrock helped me understand how different cloud services can work together to create an autonomous workflow without maintaining a traditional server.
Another important learning was the difference between frontend functionality and real backend functionality. A button that looks like it works is not enough. The complete flow needs to be verified: request → backend → AI service → database → response → UI.
I also learned the importance of handling real application states. Empty states, API failures, AI failures, duplicate requests, loading states, and deployment issues are all part of building a usable application.
Finally, this challenge changed how I think about AI applications. Instead of asking, "What can I make the AI generate?", I started asking, "What useful work can the agent do without waiting for the user?" That shift in perspective was probably the most valuable part of this challenge.
DailyMuse AI is my attempt to explore that idea through a small but practical autonomous creative experience.



Try DailyMuse AI


Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article