AWS Builder Center
Weekend Creative Agent Challenge: TECHPULSE — An Autonomous AI Technology Learning Agent

Weekend Creative Agent Challenge: TECHPULSE — An Autonomous AI Technology Learning Agent

TECHPULSE is an always-on AI-powered technology learning agent built for the AWS Weekend Challenge. It autonomously generates a new technology-learning edition using Amazon Bedrock and AWS serverless services, stores the content, and makes it available through a live web application—so a new learning experience is ready without manual user input.

AWS Certified Solutions Architect | Cloud & AI Enthusiast | AWS Builder

Weekend Creative Agent Challenge: TECHPULSE

Vision & What It Does

Technology changes every day. New architectures, AI techniques, cloud services, hardware technologies, frameworks, and engineering concepts constantly emerge. The challenge is no longer simply finding information — it is knowing what is worth learning next.
This idea led me to build TECHPULSE, an always-on AI-powered technology learning agent created for the AWS Weekend Challenge.
The challenge prompt asked us to turn a creative application into an always-on agent that makes something new on its own and has it ready when we return. Instead of building another application that waits for a user to enter a prompt, I wanted TECHPULSE to work proactively.
The core idea is simple:
What if a technology-learning experience could be created automatically every day, without the user having to ask for it?
TECHPULSE generates a structured technology-learning edition automatically. Each edition takes a technology concept and transforms it into an educational, engineering-focused story.
The generated content includes sections such as:
  • THE IDEA — What the technology really is
  • THE PROBLEM — Why it was needed
  • HOW IT WORKS — A simple explanation of its operation
  • WHY IT MATTERS — Its impact and importance
  • WHERE IT IS USED — Real-world applications
  • CONNECT THE DOTS — Connections to related technologies
  • ENGINEERING INSIGHT — A deeper technical perspective
  • THINK ABOUT IT — A question that encourages further exploration
For example, one of the editions generated by TECHPULSE focused on RISC-V, explaining its open instruction-set architecture, motivation, operation, applications, modular design, and engineering implications.
The important difference is that the content is not manually written and uploaded every day. TECHPULSE is designed to generate new creative learning output automatically through an event-driven workflow.
The user simply returns to the application and finds a new edition ready.
That is the experience I wanted to create:
The best learning tool is sometimes the one you never have to open to start working.



How I Built It

I approached TECHPULSE as an end-to-end cloud application rather than just an AI prompt.
The first step was defining the output structure. I wanted the agent to produce something more useful than a generic AI-generated paragraph, so I created a consistent learning format that combines explanation, engineering context, applications, and reflection.
The next step was connecting the AI generation process to AWS.
I used AWS Lambda as the serverless execution environment for the TECHPULSE agent. The workflow can be triggered automatically rather than requiring me to manually execute the application.
For the generative AI component, I integrated Amazon Bedrock. Bedrock provides the intelligence used to transform the selected technology into the structured TECHPULSE edition.
Once the content is generated, it needs to persist somewhere. I used Amazon S3 to store the generated editions.
I then created an Amazon API Gateway HTTP API to provide access to the generated content for the frontend.
For the user-facing application, I built a lightweight viewer using HTML, CSS, and JavaScript. The viewer presents the generated edition in a clean reading format.
Finally, I connected the project to GitHub and deployed the viewer using AWS Amplify.
The live application is now available on the web.
During development, one of the biggest challenges was connecting all the individual AWS components correctly. Creating a Lambda function or an S3 bucket independently is straightforward, but building a complete workflow requires thinking about how each service communicates with the next.
I had to work through API routes, Lambda integrations, deployment configuration, GitHub integration, and Amplify hosting before the complete system came together.
Seeing the generated RISC-V edition appear in the deployed TECHPULSE viewer was an important milestone because it demonstrated that the system was no longer just a local prototype.



AWS Services Used / Architecture Overview

TECHPULSE uses a serverless AWS architecture.

Amazon EventBridge

EventBridge provides the scheduled/event-driven trigger for the autonomous workflow.
Instead of requiring manual initiation, the system can trigger the generation process according to the configured schedule.

AWS Lambda

Lambda provides the serverless compute layer.
It runs the TECHPULSE agent workflow and coordinates the generation and storage process.

Amazon Bedrock

Amazon Bedrock provides the generative AI capability.
The agent uses the model to generate the structured technology-learning edition.

Amazon S3

Amazon S3 provides persistent storage for the generated editions.
This allows the generated content to exist independently of a single Lambda execution.

Amazon API Gateway

API Gateway provides the HTTP API layer through which the frontend can access the generated content.

AWS Amplify

AWS Amplify hosts the TECHPULSE web viewer and connects the deployed application to the GitHub repository.

GitHub

GitHub is used for source-code management and deployment integration.

Architecture

The overall workflow can be represented as:
Amazon EventBridge ↓ AWS Lambda ↓ TECHPULSE Agent ↓ Amazon Bedrock ↓ Generated Technology Edition ↓ Amazon S3 ↓ Amazon API Gateway ↓ TECHPULSE Viewer ↓ AWS Amplify
This architecture allows the application to move from scheduled generation → AI creation → persistent storage → API delivery → live presentation.
The result is an application designed to continuously produce new learning content without requiring a user to manually initiate every generation cycle.



Demonstrating the Always-On Concept

The central requirement of this challenge is not simply using AI.
It is building something that acts on its own.
TECHPULSE was designed around that principle.
A conventional AI learning application might work like:
User → Prompt → AI → Response
TECHPULSE instead follows:
Schedule/Event → Agent → AI Generation → Storage → Viewer
The creative output is the technology-learning edition itself.
This means that the system can create a new piece of educational content and have it ready for the user to discover later.
The deployed application provides the visible result of that workflow, while the AWS services provide the infrastructure that enables the automation.
This is what makes TECHPULSE an always-on creative agent rather than simply an AI-powered webpage.



What I Learned

The AWS Weekend Challenge pushed me to think differently about application development.
Before this project, it would have been easy to think of AI as simply a model that receives a prompt and returns an answer.
TECHPULSE helped me understand that the more interesting possibilities appear when AI is combined with automation, cloud infrastructure, APIs, storage, and deployment.
I gained hands-on experience with:
  • Amazon Bedrock
  • AWS Lambda
  • Amazon S3
  • Amazon EventBridge
  • Amazon API Gateway
  • AWS Amplify
  • Serverless architecture
  • REST/HTTP API integration
  • GitHub-based deployment
  • Generative AI workflows
  • Cloud application architecture
I also learned an important engineering lesson:
Building a product is not just about making individual components work. It is about making the entire system work together.
The process of connecting the agent, AI model, storage, API, and frontend gave me a much better understanding of end-to-end cloud application development.
Another important lesson was designing around autonomy.
Instead of asking, “What can the user do with this application?”, I started asking:
“What can the application do for the user without being asked?”
That change in perspective influenced the entire design of TECHPULSE.



What's Next?

TECHPULSE is currently at its first working version, but there are several directions I would like to explore next.
Future improvements could include:
  • A history of previous editions
  • Better topic diversity
  • Technology categories
  • Search across generated editions
  • Personalized learning paths
  • New-edition notifications
  • More interactive learning elements
  • Improved topic selection
  • Tracking which technologies users explore most
  • Richer explanations and examples
The long-term vision is to evolve TECHPULSE from a daily technology content generator into an autonomous learning companion for emerging technology.



Links

The deployed TECHPULSE application is publicly accessible here:
The source code for the deployed viewer is available publicly on GitHub:



Acknowledgements

I would like to thank Rajkumar Sir for introducing us to the AWS Weekend Challenge and encouraging us to explore AWS and build something of our own.
I am also grateful to Sangeeta Mam for her support and encouragement throughout the development process.
This challenge gave me the opportunity to take an idea from concept to a deployed cloud application and, more importantly, to experiment with what an always-on AI agent can do.



Final Thoughts

TECHPULSE started with one simple question:
What if technology learning could come to you automatically instead of you having to search for it?
The AWS Weekend Challenge gave me the opportunity to turn that question into a working system.
Today, TECHPULSE can generate technology-focused learning content, store it, expose it through an API, and present it through a live web application.
The first edition is already live.
The next one can be different.
And the goal is simple:
Keep technology learning moving — automatically.
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