Case study · AI news app · Product Build
NewsbitAI news app Case Study
An AI news app that ranks the day's top 10 stories and summarises each one with a link to the original, refreshed automatically every morning.
- Type
- Own product
- Role
- Design, backend, frontend, deployment
- Status
- Live at newsbit.in
- Service
- Product Build
- Last updated
The problem
Busy people don't have time to read dozens of long articles just to know what happened today. Most news apps hand you more to read, not less.
What I built
Newsbit pulls the day's headlines, ranks the top 10 and uses Groq's Llama 3.3 to write a short summary of each, with a link to the original article. A scheduled job refreshes everything at 6 AM IST, and a chat lets readers ask follow-up questions about the news.
Walkthrough
See it working
From the morning brief to asking Newsbit a question, in one minute.
What it does
The features that matter
What it actually does, one feature at a time.
Today's Brief
The day's biggest stories as short summaries, each tagged with the outlet it came from, so you can check the source before you trust the summary.
Sources you can open
A Sources list shows every outlet behind the brief, and each summary links straight to the original article.
Top Stories
Ranked story cards with a category, a headline, an AI summary, the publisher and how long ago it was published.
Explore by topic
Browse the same ranked, summarised stories by category: AI, Business, Education, Entertainment, Health, Politics, Science, Space, Sports, Technology and World.
Ask Newsbit
Ask about today's news in a chat, or start from a suggested thread such as a top-10 summary or “explain the biggest story”. Past chats are saved in a sidebar, and the app reminds you to check the original source.
Talk to it
Speak a question and Newsbit listens, looks into it and reads the answer back. Briefs and stories can also be read aloud with one tap.
Under the hood
How it works
The part nobody sees, step by step.
Fetch
A scheduled job pulls the latest headlines from a news API every morning at 6 AM IST.
De-duplicate
Articles already stored are filtered out by URL, so the same story never appears twice.
Summarise
New articles go to Llama 3.3 on Groq, which writes a short summary of each.
Store and serve
Articles and summaries are saved in PostgreSQL and served by a FastAPI backend to the Next.js frontend.
Screens
Every screen, as shipped
Captured from the live product at www.newsbit.in.
Read the news
The brief, its sources, the ranked top stories and a browse-by-topic view.
Ask it anything
A chat for follow-up questions, with suggested threads to start from.
Talk to it
Ask out loud. It listens, looks into it and speaks the answer back while the stories appear in the chat.
On mobile
The same product on a phone.
The stack
- Frontend
- Next.js · TypeScript · Tailwind CSS
- Backend
- FastAPI · SQLAlchemy (async) · APScheduler
- Data and AI
- PostgreSQL · Groq (Llama 3.3 70B)
- Delivery
- GitHub Actions · Render
The result
98%
of scheduled pipeline runs completed: 196 of the last 200, between 13 Aug and 2 Oct 2026
Source: GitHub Actions history for fetch-news.yml
Related Service
Service 04
Product Build
Need a system like Newsbit? I audit your workflows, build custom integrations, and hand over the keys in your accounts.
Start a project
Tell me what needs to change
Share your product and the work you need. I'll help you find the right scope and next step.