Interview for NL Newslinet.com, also on air on MW 1575 KHz in the Milan Area on Radio Centrale Milano

Original audio available here: https://open.spotify.com/episode/3XHR6bNv87AUtfXPCEpeSi?si=UXgty_N5RC2lXdQrjtbq7Q

Newslinet.com article in Italian avilable here:

---

Sara Beykpour (CEO, particle.news) on the future of news and publishers in the era of AI

In our previous article, we introduced an app and website that we believe represent the type of news access in the “post-SEO” era. The application, called Particle.News, leverages all the capabilities made possible by so-called large language models (LLMs): creating a personalized feed, generating an original article by drawing from various publications, showing opposing viewpoints on the same topic, allowing each reader to ask questions, and much more.

All of this is achieved through structured and contractual “scraping” of sources: in other words, not stealing content from original publications but receiving it transversally through dedicated and compensated connections with major publishers.

Newslinet had the opportunity to interview the co-founder and CEO of Particle.News, Sara Beykpour. The interview, organized and coordinated by the company’s editorial and communications manager, Ellie Smith, took place on July 30, 2025.

NL – Thank you for being here with us today. To start, a few words about you and your experience before founding Particle News.

S.B. – Thank you for having me, Marco. Before my co-founder and I started Particle, I spent most of my career at Twitter, before it became X. I was at Twitter for over a decade, working on various teams in product and engineering, including the Conversations team, Periscope, the live streaming app, Vine, the short video app, and also the advertising team and Twitter Blue, which was Twitter’s first subscription product. I spent most of my career thinking about how people connect with what’s happening in the world and how they get connected to the things they’re interested in. That foundation inspired my co-founder and me to start Particle with a similar mission: to connect people with what’s going on and help them understand what’s happening around them.

NL – I’ve been using both Vine and Periscope. At some point, it was used by several of my journalist friend in Milano to gather real time news. Whenever something was happening, they looked at it. But then that app disappeared…..

S.B. – It was a moment in time. It was the first time you could pick a place in the world, zoom in to any point on the planet, and see what was happening live from people who were broadcasting. That was a special moment that shrunk the feeling of the world—it no longer felt so big. But the novelty didn’t necessarily live on. The challenge with live broadcasting is that you need people watching at the same time. Influencers continued to thrive, but for everyday people, it didn’t always make sense as a use case. It evolved, though. Now we have influencers posting all the time, sometimes live, sometimes not. Twitch has live broadcasting, and you can still do it on Instagram and other platforms. But Periscope was the first to bring that to the masses.

NL – Back to particle.news. In a previous article on NewslineNet, we tried to explain what the platform is and how it works. Did we miss any important details?

S.B. – I did look at the article, in any case let me describe it for your readers and listeners. The goal of Particle is to help you understand news and events you’re interested in more quickly. Particle generates summaries of news across various sources and presents them in a personalized feed. You can keep personalization light, like focusing on tech or politics, or get very granular and fine-tune your interests. Our goal is to meet you where you are. If you’re busy and only have time to scroll through a few headlines, we’ve got you covered. If you have more time and want to dive deeper, Particle acts as a map to navigate different sources, ask questions, and fully understand what’s going on. People are busy, and keeping up with the news can be overwhelming, even if you’re motivated. Particle gives you the information you’re looking for in the time you have.

NL – One feature that seems particularly innovative is the “opposite sides” and “questions” features, which remind me of a time when I bought two newspapers, one leaning left and one leaning right, to see both perspectives. Can you explain these features, how they work, and how you select the opposite views?

S.B. – Your analogy of buying two newspapers from different sides is exactly what we’re trying to fulfill. Today, it’s hard to find alternative perspectives. If you see something on social media, you might wonder if that’s the only viewpoint, but finding others is a lot of work—reading comments, searching newspapers’ websites, or getting distracted by unrelated content. Most people won’t do that. Particle’s summaries are part of a “liquid content” offering that adapts to your needs. There’s a basic bullet-point summary, an “explain like I’m five” version that simplifies language, and a “who, what, where, when, and why” format for simplicity. The “opposite sides” feature takes different perspectives from various sources and organizes them into two sides, showing the back-and-forth arguments. It could be political, representing different parties, or for a movie, it might be what people liked versus what they didn’t. It organizes the story so you can understand how people have reacted.

Publishers and AI

NL – Italian publishers are struggling to survive in a world of SEO and Google dominance, and they fear artificial intelligence will make things worse. We’re writing a series of articles about these challenges. What is the sentiment among U.S. publishers you’ve spoken with? Are they as concerned, and what is your vision on this subject?

S.B. – I don’t want to speak for publishers, but we talk to them often and get a sense of their sentiment. There’s definitely fear and many questions about how this will work in the future. That sentiment has evolved over the last two years as we’ve worked on Particle. Publishers are starting to visualize future business models in an AI world where agents read content and users seek information differently. Everything is moving fast, and it’s challenging to keep up. Particle aims to be part of the solution by harmonizing between users seeking content and publishers figuring out how their content is distributed. We’re going on this journey together with our partners, taking feedback from all sides to build a model that works for everyone.

NL – A key point is the compensation of content creators. In your Medium article about expanding Particle’s publisher and visual content partnerships, you announced five new sources, including Agence France-Presse, but there’s no information on how they’re compensated. Can you give Italian publishers a general framework overview, not the amounts, but how it works?

S.B. – Our initial terms, especially with early publishers, resemble traditional content licensing and syndication deals, like those with wire services such as AFP, AP, and Reuters. That’s how we started, but we’ve made it clear this is just to get going, not necessarily the long-term model. Many publishers are forward-thinking and open to experimenting with us in this new world. We’re working with established partners and emerging marketplaces like Toollbit…

N.L. – …Tool…? Can you spell it out for us?

S.B. – It is spelled T-O-L-L-B-I-T. We established a partnership with them to develop a sustainable, profitable, interest-aligned content usage-based commercial model. Toollbit acts as an exchange between content providers and companies like Particle, handling payment flows and metering with dashboards for both sides. They’ve been an awesome partner, and we’ll send you more information about them.

NL – Is the partnership with publisher just contractual, or is there a technical side, like software or API integrations that they need to do?

S.B. – There’s a technical side because we’re interested in a encrypted and authenticated digital feed. An authenticated private API access to content for real-time, optimized access. We also offer an in-app native experience for hosted content and work with publishers to find the best way to display and interact with it. Some publishers are interested in technology we’ve created, which they can use in their newsrooms, not necessarily shown to end users but useful for their processes. For example, some newspapers might want to embed their articles with our interface, which we’re interested in supporting.

Meritocracy

NL – Are you considering extending partnerships to smaller, specialized entities? In Italy, Newslinet is small compared to traditional newspapers like La Repubblica or Corriere della Sera, but it’s influential in the radio, TV, and media sector. Are you open to niche or vertical segments?

S.B. – We’re definitely open to that. We work with a couple of smaller American publishers as well. Our content ecosystem is like a meritocracy—high-quality content rises to the top, regardless of the publisher’s name recognition or funding. If it’s a niche interest with a small audience, that’s fine; we want to get that content to the right person. Particle’s personalization makes it powerful for tailored feeds. We’re also interested in non-traditional publishers, like journalists moving to newsletters or self-published outlets, and local reporting, whether from larger conglomerates or independent stations, to bring that content into the app.

NL – Some journalists have moved from major publishers to platforms like Medium. So it’s no longer a world of big publishers but also of independent, smart “content providers” . Is getting their content as important as (or maybe more important than) dealing with big publishers?

S.B. – It’s really interesting how platforms allow individual journalists to build their own brands. We answered this already—getting quality content from these creators is as important as, and potentially more important than, working with big publishers in the future.

NL – How quickly do you think you can expand to include French, Italian, and other countries’ content? Will you do this from San Francisco, or will you establish offices elsewhere for publishers to contact?

S.B. – We’re starting from San Francisco, where our head of media partnerships is based. Even in the U.S., we travel to New York, where many publishers are headquartered. The vision for Particle is to be a place where users can access organized, summarized information on any topic, internationally. To work directly with publishers in Italy or France, we’ll eventually need a presence there. If it were up to me, it’d be today, but building user software products takes time. We launched eight months ago, and everything is moving fast, so I can’t give a specific timeline, but hopefully soon.

NL –How do you see Particle making money and being profitable?

S.B. – We want to try a subscription model first because it aligns with our goals. Ads are part of the attention economy, which makes keeping up with news difficult because everyone’s competing for your attention. Information gets buried under ads. With Particle, we want to give you time back to understand what you need and move on with your day, whether it’s about your community, celebrities, or TV shows. A subscription model aligns better with that than ads, which encourage doom scrolling. We don’t have a subscription model yet, but we’re working on it. In a world where AI agents read content, it’s an interesting question who’s seeing ads and what that model looks like. We’re paying attention to trends in the ad industry and how AI might deliver ads in the future.

NL – Let me understand something about how publisher gets compensated. Say, I subscribe for news from France, and get a feed composed of articles created based on, say, AFP and Le Monde. Would you give money back to those French publishers? I ask, given that in particle.news I’m really reading AI-generated summaries, not their original text…

S.B. – It’s early, but we’re actively thinking about it and talking to exchanges building similar models to see if we can rely on them or need to develop something in-house. Compared to two years ago, there’s been a lot of progress because everyone’s talking about this now. Of course if a reader taps through to an article, we have ways to track that engagement for compensation, but the details are still in development.

NL – Techncal question. For every article, if I request the five Ws or other features, iit seems to me the AI is generating content for me in real-time. A lot of GPU time: how do you handle the load with millions of users? What models do you use, and are you hosting them or paying for API calls?

S.B. – The output isn’t always generated in real-time. If there’s no change to the story, it’s cached. If there’s an update, it’s run again. We use all major AI models, choosing them based on their strengths for specific Particle tasks, considering cost and latency. Inference costs are high at scale, but the most expensive part is our “reality check” system, which fact-checks summaries against sources to minimize hallucinations. We use slower, more expensive models for this to ensure accuracy and nuance. That summary is then shown to users, so we don’t rerun it for each view unless there’s an update.

NL – Do you have journalists working for you, and what do they do?

S.B. – We do have journalists on staff. They do things like correct misleading or outdated summaries. For example, there was an earthquake off the coast of Russia yesterday, initially reported as 8.0. Our summary reflected that, but hours later, it was updated to 8.7 or 8.8. Our journalist updated the summary with the new information. They also handle user questions or doubts, researching and correcting summaries as needed to ensure accuracy.

NL – Any final advice for Italian publishers about the future of their business or how to relate to AI startups like Particle?

S.B. – AI companies are enabling users to consume content differently, and users are vocal about what they want. Over the past 15–25 years, as content moved online, many user needs were ignored because tailoring content to individuals was hard. AI offers an opportunity to meet those needs. Particle listens to users, and if the industry—publishers, content providers, and tech companies—listens to what users are saying, we can align incentives to help people be better informed, which benefits everyone.

SUMMARY

Sara Beykpour, co-founder of Particle News, shared her extensive experience from over a decade at Twitter, where she worked on teams like Conversations, Periscope, Vine, and Twitter Blue, shaping her mission to connect people with relevant news through Particle. The platform aims to deliver personalized, summarized news feeds that adapt to users’ time and interests, with innovative features like “opposite sides” and “explain like I’m five” to provide diverse perspectives and simplified explanations. Particle is navigating the evolving media landscape by fostering partnerships with publishers, including traditional licensing deals and emerging marketplace models like Toollbit, to create a sustainable ecosystem that balances user needs and publisher compensation.

Addressing concerns from Italian publishers about AI’s impact on their industry, Sara acknowledged the fear and rapid changes felt by U.S. publishers as well, emphasizing Particle’s commitment to being part of the solution through collaborative partnerships. While currently U.S.-focused, Particle has international ambitions and is open to working with smaller, niche publishers and independent journalists to deliver high-quality, tailored content. The company is exploring a subscription-based monetization model to prioritize user experience over ad-driven attention, with technical integrations like API access and in-app content hosting to enhance publisher collaboration.

Keywords: particle news, personalized feed, ai summaries, opposite sides, content licensing, publisher partnerships, toollbit, subscription model, niche publishers, real-time news, fact-checking, media ecosystem