-
-
Meet NewsPilot: your personal AI news agent.
-
Agent processes the request
-
final results
Inspiration
The idea for NewsPilot came from a simple problem we all have: there is too much information online.
If you want to know what happened in a specific field, you usually have to search for it yourself, open different websites, scroll through articles, and figure out which news is actually relevant. By the time you're done, you've already spent a lot of time just trying to catch up.
So we thought, why not make this easier?
Instead of searching for news yourself, what if you could just tell an agent what you're interested in and let it find the latest information for you?
That is how NewsPilot started.
What it does
NewsPilot is a voice-based news agent that helps users stay updated on the topics they care about.
The user simply starts a briefing and talks to the agent. They can ask about a specific topic or domain, such as AI, healthcare, technology, business, or anything else they want to follow.
NewsPilot then searches for the latest relevant news and presents it back to the user in a clear and simple way.
The main idea behind the project is to reduce the effort of staying informed. We didn't want to create another platform where users have to scroll through an endless feed. We wanted the agent to do the searching and organizing instead.
How we built it
We built NewsPilot around an agent-based workflow.
The agent takes the user's request, understands what they are interested in, searches for relevant and recent information, and turns the results into a news briefing.
We also focused a lot on the interaction. Instead of designing the experience around search bars and filters, we wanted users to communicate naturally with the agent. The voice interaction makes the process feel more like asking someone, "What's happening in AI today?" rather than manually searching for keywords.
After retrieving the information, NewsPilot presents the results in a clean interface so the user can quickly understand what is happening without feeling overwhelmed.
Challenges
One of the biggest challenges was making sure the project was more than just an AI chatbot that searches the web.
Getting information is one thing, but understanding what the user actually wants, finding relevant and recent results, and presenting them in a useful way is much harder.
We also had to think about how much information to show. Too little information makes the briefing incomplete, but too much brings back the same problem we were trying to solve in the first place.
Finding that balance was an important part of building NewsPilot.
What we learned
This project taught us that building an AI agent is not only about connecting an LLM to different tools.
The workflow and user experience matter just as much.
We learned that an agent becomes more useful when it has a clear job to do. In our case, that job is helping users stay informed without making them spend time searching through multiple sources.
We also learned how important it is to think from the user's perspective. Sometimes giving people more information is not the answer. Sometimes helping them find the right information is more valuable.
What's next
NewsPilot is currently focused on making it easier to get the latest news about topics you care about.
In the future, we would like to make the experience more personalized, allowing the agent to better understand the domains and topics each user follows.
The long-term idea is simple: instead of opening several apps every morning and trying to figure out what happened, you could just ask:
"NewsPilot, what did I miss?"
And let the agent do the rest.
Log in or sign up for Devpost to join the conversation.