The media industry today may not have a very favorable view of AI. The technology has already been used to replace journalists with AI-written copy, while other AI companies have been stealing journalists’ work to satisfy their chatbots’ data needs but not returning it. It brings traffic to publishers, just like search engines did. But one startup, an AI newsreader called Particle created by a former Twitter engineer, believes AI can play an important role in the media industry by helping consumers understand news and dig deeper into stories, while also finding ways to support publishers. ‘ Enterprise.
Particle, which received $4.4 million in seed funding from Kindred Ventures, Adverb Ventures and other angels, was founded last year by former Twitter senior director of product management Sara Beykpour, who worked on products such as Twitter Blue, Twitter Video and Conversations. He is the person behind the experimental app twttr. Her co-founder is Marcel Molina, a former chief engineer at Twitter and Tesla.
From a consumer perspective, the core idea of Particle is to help readers better understand news through AI technology. Particle offers a variety of clever features that allow you to approach the news in a variety of ways beyond summarizing stories into bullet points so you can catch up quickly.
But rather than simply siphoning off publishers’ work for its own purposes, Particle aims to reward publishers or even drive traffic back to news sites by prominently displaying and linking to sources right below the AI summary.
To start, Particle has partnered with select publishers to host some of their content in the app via APIs, including outlets like Reuters, AFP, and Fortune. These partners receive better positioning and their links are highlighted in gold compared to other partners.
Beta testing has already shown that readers are clicking through to the publisher’s site because of the app’s design and user interface. But now that the app is rolling out to the general public beyond news addicts, that may be changing. Over time, the company plans to introduce other ways to work with the media in addition to sending referral traffic. The team is also discussing with publishers about providing users with access to paid content in a way that suits all parties.
“Having deep partnerships and collaborations is one of the things we’re really interested in,” says Beykpour.
To help with traffic referral efforts, the article section of the app includes Big Tab targets to make it easy for readers to click through to the publisher’s site. Particles also feature journalists’ faces, and readers can follow or read more content through links to publisher profiles.
News consumers can use the app’s built-in AI tools to switch between different modes, such as “Explain it like I’m 5 years old,” to create a simplified version of a complex article or “just the facts” (or the 5W’s — who, what, when, where, why). On the go, you can listen to news summaries in languages other than English, audio summaries of stories, or personalized collections of stories. Particle can also pull out important quotes from stories and other reference links.
But two of the more interesting features have to do with how Particle leverages AI to present news from different angles and allow you to engage more deeply with the current story by asking questions.
At Particle, one of our tools called “Opposite Sides” aims to break users’ filter bubbles by presenting different perspectives on the same story. This model has been tried before by other news apps, including startups Brief and SmartNews. Unlike previous work, Particle includes a story spectrum that shows how news is being reported on both “red” and “blue”-leaning sites, how far to the left or right the news is positioned, and how big it is. A bubble representing is placed. Coverage can be on one side or the other. AI can also summarize both sides’ positions, allowing news consumers to get their own opinions on the issue.
But the killer feature of the app is its AI chatbot that lets you ask questions and get instant answers to your stories. The app includes suggested questions and questions asked by others. For example, if you’re reading about Trump’s immigration policy plan, you can ask the chatbot questions like “What are the potential legal issues with Trump’s deportation plan?” or “What is the potential cost of mass deportation?” Above all. Particle then uses AI technology to find answers and fact-check for accuracy.
“The chat feature uses OpenAI as well as our own pre- and post-processing,” Beykpour explains in an interview with TechCrunch. “If I want to find additional information on the web, I use content and search the web a bit to generate an answer.” She says Particle includes an additional step where, after an answer is generated, the AI must find supporting material that matches that answer.
Overall, the app includes technologies such as OpenAI’s GPT-4o and GPT-4o mini, Anthropic, Cohere, etc., as well as more traditional AI technologies from Google that are not LLM-based.
“We have a processing pipeline that takes relevant content, summarizes it into bullet points, headlines, sub-headlines, and does all the extraction,” she continues. “And then… Pull out quotes, links, and all kinds of related information about (the story). And we have our own algorithm for ranking, so the most important or relevant link is the first link you see, or the quote you think is most important or relevant is the first link you see.”
The company claims its technology reduces AI accuracy problems by one in 100 and the likelihood of them by one in 10,000.
Particle will also use human editors as it grows to better curate AI content and manage the homepage, she says.
The app is free to download on iOS for the time being and works on iPhone and iPad.
