Strava app - AI gist generation

AI-powered Strava integration: analyze new workouts with Amazon Bedrock and post personalized insights back to your activity feed—fully serverless and event-driven

Comment

Inspiration

As a Strava user, I appreciated apps that could populate my activity with additional info. This sparked an idea: what if an application could do more - and give AI summary of it? My project was born from the desire to create a "virtual coach" that uses AI to analyze workout data and provide unique, encouraging feedback right in the activity feed.

What it does

This project provides a fully automated, AI-powered analysis of new Strava workouts. It uses an event-driven AWS serverless architecture to listen for new activities. Once an activity is logged, a Lambda function fetches the data, sends it to Amazon Bedrock for analysis, and then posts the personalized AI-generated insights back to the user's Strava activity description.

How we built it

I built this using a modern, event-driven architecture on AWS. The backend logic is written in Python running on AWS Lambda, triggered by events from Strava's webhooks via API Gateway. For the frontend, I created a simple and clean single-page application (SPA) using HTML, CSS, and vanilla JavaScript.

Challenges we ran into

The start challenge was correctly configuring the IAM roles and permissions between API Gateway and Lambda. Bigger problems started when debugging and catching events in Amazon CloudWatch Ensuring that event payload was passed correctly through each stage of the event-driven architecture - that required careful investigation and debugging. Here Q was of much help in some specific detail revelings.

Accomplishments that we're proud of

My biggest accomplishment was successfully integrating seven different AWS services into a single, cohesive application. It proves how powerful a well-designed, event-driven architecture can be.

What = learned

I learned the immense value of a decoupled, serverless architecture. By using services like API Gateway, I could build a system that is resilient, scalable, and cost-effective, without managing a single server. It was a practical lesson in building for the modern cloud.

What's next for Strava app - AI gist generation

The current version is just the beginning. The future of this app involves building out a rich user portal. Since all workout data is saved, I envision creating a dashboard with historical progress charts, performance trend analysis, and allowing users to set goals that the AI can track and offer feedback on.

Built With

Share this project:

Updates