Inspiration
Lately, I’ve been getting more interested in keeping up with the news, but I quickly realized that reading through full articles often feels overwhelming—too much information, too scattered, and often not structured in a way that helps me get what I actually care about. I wanted something more direct, something that would let me understand the core of a news story at a glance.
That’s when I thought of the 5Ws and 1H technique—Who, What, When, Where, Why, and How—a method widely used in journalism and investigation because it efficiently captures all the essential details of any situation. It made perfect sense to apply this structure to summarize news articles: it’s fast, focused, and familiar.
So the inspiration really came from a personal need for clarity and relevance in a world flooded with information, and a desire to reformat the news experience into something smarter and more accessible.
What it does
This browser extension helps you quickly understand any online news article by summarizing it in a clean, structured format using the 5Ws and 1H technique:
- What happened
- Who was involved
- When it happened
- Where it took place
- Why it matters
- How it unfolded
Instead of digging through long paragraphs, you get all the key information right away. The Who and Where sections come as a list of names and places you can explore further—each one can be looked up on Wikipedia for more context.
If you prefer something even shorter, the extension also provides a one-paragraph summary.
To go deeper, it gives you a list of related articles and sources, so you can read more from different perspectives—all automatically gathered and summarized using Perplexity AI’s Sonar model.
In short, it turns long articles into quick, useful insights—perfect for staying informed without feeling overwhelmed.
How we built it
We built the extension with speed, scalability, and ease of use in mind. To keep everything fast and responsive, we used a serverless backend, which handles the summarization process on demand.
To make the most of Perplexity AI’s Sonar model, we used Genkit, an AI application framework that helps structure interactions with language models. Since Sonar was new to the ecosystem, we went a step further and developed a custom plugin to integrate Sonar into Genkit—making it easier for us (and now other developers) to use Sonar within Genkit projects. You can find the plugin's repository link below
Here’s how it works behind the scenes:
- When a user clicks “Summarize” on a news article, we first check if it has already been summarized before—using the article’s canonical URL.
- If it’s already in our database, we return the saved summary instantly.
- If not, we generate a summary in real time. We send a carefully crafted prompt to Sonar, which verifies that the page is a real news article, finds related articles across the web, and creates a structured summary using the 5Ws and 1H format.
- The result is then saved for future use to make future requests faster.
This setup allows us to deliver smart, reliable summaries quickly, while keeping everything efficient and reusable.
Challenges we ran into
One of the biggest challenges was how new the Sonar model is—there wasn’t an official integration or much tooling available yet. To make it work smoothly with our development setup, we had to build our own plugin to connect Sonar to Genkit, the AI application framework we used. This took time, especially since we had to understand both systems deeply and make sure the integration was stable, flexible, and ready for future use.
Another major challenge was prompt engineering. We needed a single, smart prompt that could do several things at once:
- Validate that the provided URL is actually a news article
- Search for and read related articles across the web
- Extract key information in a structured 5Ws and 1H format
- Summarize it again in a short paragraph
- Provide a clean, machine-readable format we can use in the extension
- Getting all of that to work in one flow—and do it consistently—took a lot of iteration and testing.
It was a mix of technical unknowns and creative problem-solving, but in the end, those challenges helped shape a better, more robust tool.
Accomplishments that we're proud of
One of the accomplishments we’re most proud of is developing a new plugin that brings Perplexity AI’s Sonar model into the Genkit ecosystem. Since Sonar is still new, there was no existing way to use it easily within modern AI app frameworks like Genkit—so we built the missing piece ourselves.
What's next for The5Ws1H
In addition to pushing the first final release of the extension with In-App Purchase, here's some future features we want to add:
1 - Custom News Feeds with Summarized Headlines Users will be able to follow specific topics or sources, and get daily digests with 5Ws1H-style summaries—turning news tracking into a quick daily habit.
2 - AI-Powered Comparison View For major stories, we plan to let users compare how different outlets report the same event—showing multiple 5Ws1H summaries side-by-side for a more balanced perspective.
3 - Mobile Companion App A mobile app version that lets users save articles for later and receive bite-sized summaries on the go—great for commuting or quick catch-ups.
4 - Multilingual Summaries We're exploring multilingual support to bring 5Ws1H summaries to non-English news content, breaking down language barriers in global news.
5 - User Feedback Loop for Summary Quality Users will be able to rate summaries, flag inaccuracies, or suggest improvements—helping us continuously refine the output with real-world feedback.
And much more...
Built With
- angular.js
- firebase
- genkit
- sonar
Log in or sign up for Devpost to join the conversation.