Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
Particle & The AI News Landscape particle.news
Recorded: June 20, 2025 Duration: 1:09:51Space Recording
Short Summary
In a recent discussion, Sarah Bakepore, co-founder of Particle News, unveiled their innovative app aimed at revolutionizing news consumption through AI-driven summaries and personalized feeds. The conversation highlighted the app's launch, strategic partnerships, and successful fundraising efforts, reflecting a growing trend in the media landscape towards technology-driven solutions.
Full Transcription
Thank you. Thank you. Hello everybody, thanks for coming to Spaces.
Usually I do these spaces while doing chores, and that's no different today.
I have this big pile of laundry in front of me, which is going to take forever to do because my right hand is in a splint.
So I'm trying to be folding laundry one-handed while doing the spaces to be efficient.
So very exciting. But today
we have a special guest to my laundry folding spaces. We have Sarah here from Particle News
to talk about AI and journalism and founder life and things like that. So welcome, Sarah.
Hey, thanks for having me, Billy. Everyone can,
can you hear me okay? Yeah, I can hear you okay. So I'm assuming everyone can.
But yeah, if anybody wants to put up hearts and whatever emojis to show that you can hear,
that would be appreciated for our own sanity. Yeah. and so I started recently working with Particle News
because I think, you know, the news is pretty messed up
and it's really hard to get factual information.
So I like the idea of getting to see a lot of different sources
and what those different sources are saying at once.
And, you know, I'm not just getting information from one space because kind of no matter what news site you go on it seems like the article you get is going to be pretty biased.
And then eventually bring some people up after our kind of like introductions and we can have discussions about news or whatever.
It'll probably go in some weird directions just because that's what spaces tend to do.
But yeah, did you want to introduce yourself to everyone, Sarah?
Yeah, awesome. And thanks for the intro.
Yeah, I'm Sarah Bakepore. I'm the co-founder and CEO of Particle News. You can download it at particle.news
or you can go to the app store and just search for Particle News and you'll find it there.
And yeah, we basically started this app with the intention of making it easier for, you know,
regular people to keep up with what's going on because we are in an age of unprecedented
information overload and it's too much work to figure out what's going on and to get balanced perspectives.
And as a result, you either just are uninformed or, you know, you rely on algorithms to kind of get one side of a story.
Or, you know, you just rely on that one friend who knows everything that's going on to text you whenever something is happening.
So what Particle does is it uses AI to read all of the news for you,
and it summarizes it for you in a personalized feed,
and so you can tune it to your preferences,
and we let you kind of consume the news in a way that works for you and your life.
So a phrase that people have been using lately, and I've just started to understand what it is, is liquid content.
So I guess in an AI age, the idea is like content is liquid and it can transform to how you want it to be.
And I think that's something that we embody at Particle.
So, you know, you could get your new summary, your new summary just as we've created it for you,
or you can transform that summary into explain like I'm five,
so really simplify the story, use simple language,
assume I have no context, or make it even shorter,
just give me the five Ws, the who, what, where, when, and why.
Just give me kind of like the quick hit of that information. Or show me the five Ws, the who, what, where, when, and why. Just give me kind of like
the quick hit of that information. Or show me the different perspectives of this story. So what are
the different opposite sides and the arguments for each side? Things like that. So, you know,
we've been able to, I think, create an app that starts to kind of meet people where they are
with their relationship with the news. And because we're, you know,
grabbing news and reading news from so many different sources, it helps kind of
keep your perspective balanced and, you know, cut through all that, all the noise and with
as little work as possible, of course. Yeah, that's what's great about AI is you can let
some computer do the algorithm for you.
But that brings up an interesting thought I was having.
So you mentioned that a lot of times we get our news from sources from the algorithm, right?
And the algorithm will kind of like shoehorn the news that we start to click on or whatever.
So like, for example, if I'm watching YouTube and I start
picking specific videos, it'll start to go down a rabbit hole and get even narrower and narrower
where it tries to like give me what it thinks that I want. When you made a particle, was it
something that you wanted to allow people? like if they do want kind of one
sided news, they can have it? Or was it something more intentional of like, let's try to get rid of
that? Yeah, and I think that's a really good question. And just for some background for,
you know, folks here who were meeting me for the first time. I spent over a decade at Twitter
working on various parts of Twitter Now X.
It was Twitter when I was there.
You know, thinking about how we connect people
to the stuff that they care about
and how those algorithms work.
And of course saw firsthand,
like what it means to have an algorithm
look at your behavior and kind of take you into some of those directions
that you're talking about. So we've made some deliberate decisions with Particle on how we
think about the algorithm that we present to you. And in particular, you can choose the topics you
want to see and you can fine tune or like, you can tune those topics to be very, very specific and granular, or you can go
broad and say, you know, you're interested in, in, in things that are much higher level politics,
entertainment, whatever, but we don't let you choose the news sources that you see. So it's
pretty difficult to get into a, um, uh, into a bubble or an echo chamber where you're only getting one perspective.
We'll always show you all of the perspectives that we have for any given story.
And that, I think, will help a lot so that you don't just continuously see one-sided news.
And that goes for both sides, right?
Like, we don't want you to get stuck in a
left echo chamber or a right echo chamber. We want to make sure that you're aware of
all the perspectives.
Yeah, I think that's actually a pretty elegant solution because, yeah, it just seems like,
for example, I watch a lot of, how do you put it, like, this particular movie is bad.
That seems to be like a general topic in my YouTube
where I start going down that hill.
And that seems to always go into this right echo chamber
at some point where it's just like,
it's bad because women are bad and they're taking over.
I'm like, no, that's not.
I want to hear people shit on Game of Thrones
because the ending made me so mad.
It's been, what, like eight years or something.
I'm still mad about it.
So I will watch any game of thrones ending sucks but like it just seems to always tail toward
this kind of like narrative that i'm like that's i don't think i don't jive with that narrative
it's not exact it seems to be this like kind of echo chamber narrative and there's a lot of people
giving that narrative so yeah i think it'd be nice to like if i just like can i
see all the game of thrones is bad stuff but like varied perspectives i just want to just you know
taking all the game of thrones as bad uh content that i can yeah yeah um this reminds me of like
um there is a stand-up comedy um by ronnie chang i don't know if you know he's like on the daily show and stuff
he has this quote that's like really funny where he's like you know guys just go to youtube to try
to figure out how to you know use the right posture for their for their curls and you know
they just want to like get buff or whatever and two weeks later they're storming the capital um
it's i mean i've not delivered it the way he did but it's it's pretty funny no and it's
it's true like you go you try to become an expert in something or learn more about something on some
of these algorithms the next thing you know it's like you're in weirdo land and you're like
yeah but i can see that fringes of the internet yeah because like i can see that in kind of the
reverse where like i listen to podcasts about video games because i like video
games and like it just seems like the video game podcast is like la la la and by the way black
lives matter and blah blah it's really important like what shut up i think like that's important
to some people like cool whatever but like i want to hear about video games and what video games are cool. And I want to get into this like fucking, like
every random thing is this
the culture
war. I'm like, I don't care. I want to play
a video game. I want to see if that's good. I want to see if I
should waste my money on that.
And not be as much of a waste of money.
Yeah, like if you go too
deep on anything on YouTube, you start
to get into brainwashing territory.
It's really weird. It's really weird. It's terrifying, actually.
Yeah, yeah. And I guess that's, that's a little bit of everything. But what was, what was like
your, like, what made you decide you wanted to start this company? Like, did you have like a
founding moment where you're like, trying to find some information? You're like, oh, this is bad.
And inspiration came? Or was it kind of a longer, longer thing? You know, it's, it's interesting because I had,
I had left Twitter now X at the end of 2021 after, you know, over a decade there. And I thought at
the time, I'm like, okay, I'm kind of done with like social and news and, and, and that while I
was Twitter, while I was at Twitter, I worked, I worked on core Twitter, but I also worked on Vine and Periscope.
And actually, this product spaces that we're on right now is built on Periscope stuff that I
worked on. And I didn't think I was going to kind of go back. And I left Twitter and I was like,
you know, the company is good, you know, did put my time in there and I'm ready to kind of go back. And I left Twitter and I was like, you know, the company is good. You know, did, did, you know, put my time in there and I'm, I'm ready to kind of move on to do something else.
But kind of a few things happened at the same time. One was Elon had taken over the company.
And so there was this like question for me of like, okay, well, you know, what,
what's the direction of twitter now and
like you know where is it going to go and another thing was that chat gpt had just come out and it
felt to me like there was an opportunity now um to try some new things and maybe do some things that
i'd like you know maybe always wanted to do while i was at twitter but never really had the opportunity
to do um and so my co-founder marcel and and I, and a few other people, actually, we spent a bunch
of time brainstorming around that time.
And one of the things that became obvious that AI could help with was the idea of summarizing
the news.
When we were thinking about some of the challenges that Twitter and other
social media platforms have, oftentimes you end up in these conversations or you read these
conversations where people are just talking past each other, right? And it's like they're entrenched
in their views, they're talking past each other, and no amount of like, did you even read the
article is going to help make that conversation more productive.
And we thought, well, what if it could just be easier to read the article?
What if it could just be easier to like kind of catch up on like what the other side of that conversation might be referring to?
You know, like just really lower the barrier, make it less work.
And then also start to have a conversation based on a shared set of facts.
And that's really what the particle summary tries to get at.
It's not trying to tell you what to think or how to think.
All of that comes in in other layers.
There's quotes, and you can tap into the articles.
You can kind of go deeper and get more information.
You can see links to tweets and Instagrams and whatever.
But the summary itself tries to be a reflection of like, okay, based on everyone who has reported on this, like, what are the agreed upon facts here? And like, let's start there. And maybe that
can help kind of the baseline of a conversation. So that's kind of where that all began. And it kind of evolved into into the app that it is today.
I see. Is this the first time you've founded a company or done that kind of thing?
It is the first time. I've had like some startup experience, you know, working on smaller teams like Periscope and Vine. And I worked at a small startup in the early 2013, 2014 period called
Secret. I don't know if any of you remember that, but it was like an anonymous social network.
Wait, is it, what kind of, Secret sounds salacious. Was it like you can message someone
that you're like, that you like them and then they can figure out who it is type thing?
It was like anonymous.
It was kind of like Whisper or Yammer or Blind, where it's like an anonymous social network.
Oh, I see.
You uploaded your contact list, so you could tell if the post was from someone you knew, but you didn't know who it was.
Or you could tell if it was was from someone you knew but you didn't know who it was or you could tell if it was like from like a friend of a friend it was like one of those apps that just kind of like exploded and died within the course of months or whatever it was
a lot of fun um but uh yeah i've had a lot of like that startup experience but never um founded a
company until now i see what uh i guess, like, what's been your founder life experience
or not experience, like, what do you think of it right now?
And is it, you know, I'm imagining just from making Dogecoin
as a accidental founder and dealing with all of that,
that there's times it can be completely overwhelming
or there's a lot of things you have to do
that you don't expect that you have to do or,
or stuff like that.
Yeah. I mean, you probably had like just an insane journey. Um, but, uh, uh,
it's, it's, it's a lot of rollercoaster. Um,
and I think a lot of people say that it's a lot of rollercoaster up and down,
um, you know, highs and lows.
But we've got a really great team. And so it just it makes everything like, you know, fun.
We're having fun while we're doing this.
And we think we're working on something that we all believe in.
So that really helps kind of motivate and keep us going.
And I think, you know, working with people that you like is like a really important,
it's like, it's really important. Life is too short to work with people that you don't like.
So, you know, we've got, we've got an awesome team and they're, they're all really, really
talented. So that makes, that makes some of the challenges of being a founder kind of a lot easier.
Yeah, that's good to hear. I think I've worked in like large companies
and smaller companies and then very small company.
And I think the smallest companies
were the most fun in some ways
because you were extremely directly involved
in whether your company lives or dies.
It's like, ah, if I don't ship this,
it's probably not good.
I've kind of actually got to choose this.
But it was also like, yeah.
Our pressure is real.
Our pressure is super real.
But I don't know.
It's like, I remember we had kind of a late crew in one of the startups I was on.
And those basically all became my closest friends because we were there until like 2 a.m. on so many days.
But, you know, you can't actually straight work until 2 a.m. So it was kind of like, it felt like college a.m. on so many days but you know you can't actually straight work until 2 a.m. so it's kind
of like it's not like college almost he's like hanging out still and doing papers uh equivalent
yeah yeah um yeah I don't have the uh uh the like physical and mental fortitude to like pull
all-nighters the way I did in college college yeah no like I don't know how I did
anything like like I don't what I don't what I really don't understand this is a complete aside
but I really don't understand is like I ate like complete garbage growing up like I just say like
my mom would buy like Burger King for the week and then put like seven cheeseburgers in the fridge because they
were like 39 cents or whatever and then like you know by day five six I'm eating like six day old
McDonald's and Burger King cheeseburgers that were been in the fridge like it just in the paper
and I like lived like I don't understand like our bodies are a miracle. But nowadays I'm like, oh, I need to eat this low FODMAP diet.
If I eat too much onions, I get a stomachache, me, me.
I'm like, oh, my gosh, this is pathetic.
Yeah.
No, totally.
I, you know, I love ramen, and I definitely had a lot of ramen growing up,
and I definitely don't eat as much of it now.
But I can't say no
to an instant ramen if someone's making one. But I was like looking up, you know, really with
Chachipiti, I feel like Chachipiti has made like a big difference in my like, is this healthy? You
know, I just like kind of like ask it all the time. And it's basically like every single ingredient
in instant ramen is like terrible for you. It like could not be worse.
Well, it has so much. So did my friend., my friend, I remember, uh, one of my buddies, this is in
middle school. So I used to hang or in the mornings, um, my mom went to work and she would
drop me off on the way to work to his place. And then we would walk, uh, to work after that or
walk to school after that. But, uh, we had like two hours to like fool around. And like in that
time we drank, uh drank his dad always made
coffee so we like drank the leftover coffee with like a ton of uh it was like this cream creamer
cinnamon hazelnut we just like make it like super sugar whatever and we'd eat so much of the like
march and instant noodle stuff the one that you just put in the microwave yeah and so we'd eat
like you know a couple of those and then a bunch of coffee in the morning.
And then I remember one day he told me he went to like a doctor as a, you know, seventh grader.
And they told him his like blood pressure was really, really high.
Oh, my God.
Your blood pressure is like incredibly high.
It's like maybe it's the the four cup noodles
is like now now it's just like yeah you think but back then it's like maybe yeah the idea that
you know us we could get that sick from from you know junk food every morning was like
you know so this this is when parents aren't around and the kids are just left to their own devices. We don't make the best decisions.
Yep.
Do you have like any advice then for if someone has an idea and is interested and just wants to take a chance to start in the company or starting the company of their dreams?
Sure.
Are you asking for a friend or are you thinking about that at some point?
Well, I'm very lazy.
So I'm sure that's not a good trait of any founder.
But it's kind of, you know, it's always on the back of my head of like,
I would like to do this.
And then I kind of lose interest when I realize that it takes work and effort.
But yeah.
So, I mean, for the right, you know, for the right product or for the right idea i'm sure you you
would have the motivation um but uh i mean advice to someone who's like you know getting into that
today and i'm certainly by no means an expert here but you're an expert you've done it. Yeah, fair enough. The landscape today is like so different than, you know, five, ten years ago.
And, you know, what you're able to accomplish as a, you know, solo founder or small co-founding team.
Uh-oh, did we lose you or did we lose me?
Shoot. Can we lose me? Shoot.
Can people hear me?
Let me check something.
Internetworks.
Seems like.
No, I think there's a better time to make leap oh sorry yeah i was my stuff was messed
up it seems like it came back question mark i'm not sure oh i don't know can you hear me yeah i
can hear you okay great anyway go ahead um yeah i mean there's just like no better time to do it.
It's like kind of maybe a new kind of golden age of making your ideas reality
at a very, very low effort, low cost
compared to how it used to be.
Is that, you think, in large part because of AI?
In large part because of AI.
Also, if you're starting a company, you know, and if you're in Silicon Valley, the kind of like support that you can get here, just like in terms of, not just in terms of like hiring and all that kind of stuff, but in terms of resources that
are available to help you on the founder journey.
They're so rich here in San Francisco and Silicon Valley.
It feels like we're in the primordial ooze stage again of a resurgence of just a host
of activity and excitement around around new companies so
it's uh it's a good time it's fun yeah i remember all the the different uh kind of
boom cycles i guess um i mean i i wasn't i was uh pretty young for like the y2k stuff and whatever
um and the first stuff on the internet but so i just saw it from afar. But I remember when crypto was really huge in 2016 or 2017,
and a lot of companies came up from that.
And then in, I guess it was like 2013, 2014,
when Internet of Things was supposed to be the next big thing.
And there's a bunch of companies around that.
And it seems like AI is the kind of new central point.
But for each of them, I feel like things have more legs or less legs.
And AI seems like, or what we're calling AI, which I guess is like generative LM type thing,
seems like it has a lot of utility that we're trying to figure out.
out and it's moving at kind of an incredible pace. So I guess I'm wondering what your
And it's moving at kind of an incredible pace.
development experience, you were like mostly back end in the beginning of your career and then
have been doing more and more or what would you call that? I've actually been more fun end than anything else so I started off doing kind of like web um uh web internal tools type type stuff
um and then um I moved to Android I want to say around 2011 and I spent uh I don't know how many
years maybe you did the beginning of Android when yeah couldn't even just be I do Android
development so I remember back then it's like they couldn't even decide if it was like like our activity is good do you want everything to be an
activity or this is maybe esoteric for our audience but it was it was a weird time they
didn't have any like um google was just like yeah i don't know do whatever yeah we have no guidelines
exactly i mean i think what was cool about that
is that a really cool um community kind of built out of out of that time but that's true um yeah i
did i worked on twitter for android vine for android and secret for android when i you know
worked at that startup and then periscope for android uh worked on all of those uh all of those
apps so if you're on an android phone and you're listening to this space right now,
there may be a couple of lines of code that I wrote that has lasted from here.
But I've always been more of a product engineer,
and thinking about that user experience and how people
interact with the product and the emotion
and the relationship that people have with their UIs.
That's cool.
Yeah, I like that space too just because I do mobile development
for the most part and most recently has been Android.
But I like that space because it's very easy for me
to see the result of something that I did and be like oh is
this good is this work is it usable versus like this kind of it's kind of hard to use it this
kind of stinks so I like being in that space rather than more abstract kind of like back
indie stuff where I'm just like I don't even really know uh yeah if this was valuable or not
um but that's for me. I greatly appreciate really good engineers
who can do stuff like that, though.
Oh, I was literally just about to say the same thing.
Like, I totally have a ton of respect
for really talented engineers that can do that.
Yeah, when they make beautiful APIs,
you just kind of call them to get all the stuff back
that you could possibly want.
You're like, yes, thank you so much.
Yeah, exactly, exactly. i i'm just wondering because you know i've been doing
uh ai help to development not really vibe coding but like partial vibe coding or whatever with
cursor uh i was wondering what like when you code currently or if you uh you know i'm assuming
you've done a lot of code with uh with particle but when you do it, do you have a lot of AI help
or is it like meh?
Like it's too weird state right now for that.
So I haven't done a ton on Particle.
I've done like a very small amount
and mostly in the earlier days of the company.
But before Cursor became like more mainstream, more mainstream, just using ChatGPT to paste
your code and be like, why is this wrong? Or how do we do this? Or whatever was definitely
part of my flow. I worked in SwiftUI, doing some basic things with ParticleApp in the early days.
And I'd never worked in iOS or Swift or
SwiftUI before. So that was critical. I don't think I would have been able to do it without AI.
I have used Cursor to vibe code some things that I need. Not all of it is committable.
Sometimes I just Cursor a script that I need like do something, but that's not like it's not like an enduring, you know, not enduring code that can actually be looked at by somebody else or committed to any repo.
But I think like folks on the team, you know, use it when it makes sense.
with, like, I think Copilot has also been, like, a tool that has been just as good, if not better,
than, you know, using cursor in some cases, just depending on the tasks that you're doing.
But yeah, I think the team embraces that stuff.
Yeah, I enjoy, I guess, like, in general, I think it has helped me.
But at the same time, it is, like, silly a lot of times.
I remember, like, I was like, hey, can you refactor some of this stuff into objects and just make it, like, a little more split out since it just put everything in one file?
And it was like, all right, and, like, did all this stuff.
And then at the end, nothing compiled and, like, everything was broken. And it's like, all right, well, then maybe it was like all right and like did all this stuff and then at the end nothing compiled and like everything was broken and it's like all right well
then maybe it was a little too complex for you or there's there's a lot of these um
how would you say like you're just like hey this thing doesn't work and it's like oh i see the
problem that was not the problem yeah it's like does something else stupid it's like well no that not only was that not the problem uh that that didn't solve any problem and it made new problems
like i don't know and like i i think like i'm just imagining the ai is just told like more explicitly
like say oh i think i found the problem before like doing whatever it does so like it's this
weird thing of like how language
works where it's like, it says something in English. So you like trust that part, but then
it like does all this crap and you're like, that's not even, yeah. Yeah. That's, that's not it.
But yeah, it has its uses like for certain tasks. And I think the, you know, as you,
I think there are like engineers that can make a really, really good use of using Cursor to help them in what they're already doing.
And like I said, Copilot and other autocomplete tools, once you already know what you're doing, can really help speed you up.
But I think the big companies, they're putting all of their money into this.
So I expect it's all going to just get better. Yeah, I hope so. Or get more something. I don't know. Maybe the, the UIs will look better, but the actual result won't be any better.
Like who knows what this, like, I always think with, with tech, it feels like at a certain point,
it just like the, the S curve happens and then the S curve levels out. Right. And then it feels like at a certain point it just like the the s curve happens and then it
the s curve levels out right and then it's like oh well now we're kind of close to how good it
can get and it's hard to know at that point it's definitely hard to know and we might it might need
to be like another wave of you know yeah advances here before it can get like that much better but
we'll see yeah it reminds me of like
VR because like VR every time they come up with something every x year it's like whoa this is
super cool and then I don't know it's like then you start to see where it doesn't work out right
and then it just kind of falters out and then x years pass and you're like oh my gosh this is so
crazy this is so cool I yeah that that is exactly my experience with VR.
It's like really cool for like a day.
And then you're like, cool, call me again in about two and a half years and we'll do this again.
Yeah, I guess to bring AI back to particle a bit, one of the things I was thinking about was what ai like what ai has some issues
with and what ai is really good at uh at its current iteration and i think personally i think
ai has some issues in being very like overconfident and just saying stuff that's not necessarily true
or like these kind of things but what AI is really good at is summarizing.
And one of the things I did,
like I write articles for a sub stack every once in a while.
And I would put my article in,
you know,
chat GPT or one of the AI tools and say,
you know,
like,
what is this about?
And what I was really impressed by was it could figure out when I was being
sarcastic.
And I was like, yes, like people on on on x can't even do that like they get mad at stuff that's like very obviously sarcastic but the ai could be like oh this is a satirical oracle about blah blah I'm
like thank you like if even ai can figure it out then like even the dumbest x user can just
dump this into ai and decide if it's
something they should take seriously or not. So I guess in terms of your news app, AI seems
mostly integrated in maybe the curation algorithm, but also in summarizing and stuff like that.
Yeah. We have AI kind of up and down the entire thing.
AI is used for summarizing,
it's used for checking the factuality of everything that it has written,
it's used for topic extraction,
entity extraction, and then understanding what is the story about.
There's embeddings and stuff that have been extracted so that we understand how similar two things are.
Those all use AI, not all of it is necessarily generative AI when you think of LLMs and chatbots and stuff,
but there are AI components basically up and down the stack. So we use a lot.
And it is very good at summarizing some of the challenges that we've had that, you know, we actively work to mitigate are around, you know, there is an accepted hallucination rate.
So sometimes it does hallucinate.
And so we have processes in place that, reduce the likelihood of that hallucination by, you know, every like bullet point that is written about a story is is is like checked and then double checked to ensure that it's double checked by the AI itself.
So it's like, hey, like proofread this thing that you just generated type thing.
thing that you just generated type thing yeah basically like you ask in another call you say
like hey um this is some output and here's some input is that output like generate is that output
supported by by this input and if so like explain and cite your sources and we ask it to do a lot
of reasoning yeah and that is like there's a lot of work there's like a whole pipeline that goes
into making sure that the particle stories are correct according to the source material.
It's like the most expensive part of our process, but it's obviously a very important one for a news app.
And, you know, we're continuously, you know, making it even better and better and, you know, trying to lower those error rates and
stuff. Yeah, that's interesting. So just, this is more maybe for my own information, but when you
work with AI to, I guess, help part of your algorithm or even be most of your algorithm,
if something's not working right, it's like you end up tweaking prompts. Is that how it works?
A lot of the time it's tweaking prompts,
but there's a lot of glue in between them.
And so, you know, there's other parts that can be updated
and making sure that the prompt has all the right context
or, you know, things like that
before it actually runs the completion.
it actually runs the completion.
But yeah, there is a lot of prompting.
There's a lot of guiding and gentle nudges in the right direction.
I think when we first started Particle a couple of years ago, we needed a lot more tricks like that to get it to do what you wanted it to do.
But now, a basic thing that we do that works very well is you tell in the prompt, you tell the AI how it's going to be evaluated.
And so you say, okay, here's the problem that we want you to solve, or the thing that
we want you to do, the task we want you to run, you're going to be evaluated on these
things.
So, you know, make sure that you meet these things.
Those could be formatting.
That's cool.
You give it like a score later.
It's like, here's how well you did to like get to this and we're aiming for this particular type
thing.
Exactly.
And then we actually run those evaluations after.
We call those evals.
The industry word is evals.
And we measure how well and how often it's passing all of those evals and when it fails.
And if you make a prompt change and you know all your
evals fail then you know you need to change it again until it kind of passes all those tests
but it does help to tell it how it is going to be evaluated it'll perform better if you tell it
it's kind of like you know you if you're giving someone like a like a test like an actual test
um like an exam of some kind but they don't know how they're going to be evaluated,
they're not going to perform as well.
That's true.
But if you tell them,
hey, you're going to be evaluated on your brevity
and your succinctness
and your use of capital letters
and whatever it is,
it'll perform better.
That seems like really good advice
for anyone using AI for any purpose,
I guess.
cause I can,
you know,
ask it about actually very recently when I,
when I hurt my hand,
I was trying to explain to it.
What happened?
You mentioned at the beginning.
it was,
it was stupid,
pathetic freak accident.
So like,
I literally was like,
I was in the middle of cooking and I wanted to get a plate and
then my cat decided that he would just like jump up into the cupboard while I was trying
to get plates.
And I was like trying to get them out.
And then I don't like, I wasn't fully paying attention to what happened, but one of the
plates fell and it's this like clay plate that like, uh, I hate this plate and it's
now in the garbage because, uh, whatever, but like, it's this plate and it's now in the garbage because uh whatever but like this is
really heavy like clay plate and like hit the counter or something like shattered and then
like one of the the pretty uh sharp edges just kind of like jammed into my my hand and i didn't
even feel it right away i was just like like ah because I'm all mad at the cat and
I like looked at my hand I'm like oh that's gnarly it's like it was like deeper than I've
ever seen in a like I saw some like white stuff like under the cut and I was like ah so I was
like I need to go to the ER um oh no yeah so you had to go to the ER for it yeah well I mean
if I thought I saw bone or whatever the heck that white stuff was, I was just like, I don't know what that is.
I don't think I need stitches.
And yeah, I had like a wad of paper just like put over my wound because I like didn't bandage it up nicely or anything.
I was in such a hurry.
And I drove to the er mostly with my knee and then i was like wandering around the hospital because i couldn't find where the er
was wandering around with like bleeding all over the place it was pretty funny uh but yeah then i
just needed a bunch of stitches and they're like hey you like cut your tendon i'm like well that
sucks so now i'm in this cool cast or not cast this This is like a splint, but that was, that was fun. But
yeah, like, uh, during that time I was talking to, uh, chat GPT because I was like, uh, like,
oh no, like, I don't know how bad this is. And it was like freaking me out. Cause it was like,
like, it's an emergency, blah, blah, blah. Like get to the ER immediately, blah, blah, blah. I was
like freaking out. I'm like, oh God. And then when, blah, blah, blah, like, get to the ER immediately, blah, blah, blah. And I was, like, freaking out. I'm like, oh, God. And then when I was asking about, like, the cut
tendon, it was like, oh, don't move it at all. If you move it up, like, it could snap, and then you
won't be able to use your finger, and you need surgery, and blah, blah, blah. And I was, like,
freaking out. So that actually, like, you know, got me back to my old days of Googling
medical stuff, and it always saying that i had cancer
right it's like oh i have a stomach egg he's like well that's because you have stomach cancer like
thanks google yeah so yeah yeah it was like freaking me out but like maybe if i gave it
some more like if i you know i could only type with my left hand anyway but if i could type a
little longer and gave it some parameters for the end it's like you would be graded on not freaking out a potential hypochondriac or whatever it could be like oh yeah
well here are the options or you know just be a little more uh but you know at the same time this
medical stuff i don't know if they have worries about because like now i feel like whenever i
ask this stuff it's like you should ask your doctor or like it's good to follow up with the doctor so maybe there's some
worried about lawsuits or something yeah i'm sure there's like liability issues same with like you
know financial advice and all that kind of jadgbt told me to short this and now i'm you know yeah
yeah so they probably had to put like extra like if or extra prompting whatever like if if
someone asks about this make sure to say like like yeah this might destroy your life but uh
don't don't take my word for it yeah all right cool um i will open this up this chat oh wait
there's no one is asking for uh uh to come up on stage but if anyone wants to
come up on stage and chat with us talk about ai or if you have any questions or just want to shoot
the crap um i am like oh go ahead oh well while you're doing that there's some questions in the
in the tweets oh yeah yeah uh by the way you still call it tweets is that just a habit
oh god what do you what do you say what do you say no i so to most people like especially outside
people i still say tweets um and actually i've made this point many times that tweet is actually
descriptive of what it is and post is not because you have to say like a post on x you can't say just post because that's everything so
uh in my opinion that was a language downgrade to start calling it posts it's like a worse word
it's a worse word nobody wants to call it post it's a tweet it's a tweet and if it's on threads
then it's like a threads tweet. A threads tweet. I see, I see.
That's how I am.
Yeah, we got a couple people that like to come in.
But yeah, if you had some question that was on the
thread that you want to read.
Yeah, someone asked, Particle News not available on Android.
Not yet.
We only launched on iOS towards the end of last year.
So I appreciate that there are folks on Android who want to get it and we will get to it.
But I have no timeline to share on that right now, unfortunately.
But I feel your pain.
In the meantime, if you do have an iPad, we have a pretty cool iPad app.
So maybe you can give it a try there.
Someone asked, has the algorithm been consistently...
I don't know this word.
Let me see if I can see it.
From StarShiningDoge.
Has the algorithm been consistently temp-rugging a speaker when the majority of audience members get distracted by other temps?
Rugging is what people say when a space cuts out, like the space got rugged.
Oh, I see.
Just saying like it hasn't been like crapping us out purposely for algorithmic reasons.
But I think it's probably not listening to us and deciding based on what people are saying or doing that it will rug certain users.
I think it's just something I don't understand, some packet loss or something.
Yeah, yeah, yeah.
Okay, cool.
And then one more question.
Not having a specific background in news and media,
how do you pick up the domain knowledge to navigate the news media space?
So, you know, there's definitely been a lot of learning,
but also, like I said, having worked at Twitter for a long time time just being kind of at the edge of thinking about how people get connected with
the news um you know it's just it's it's i don't have like a like academic background in news but
i have some like lived experience uh from working at twitter and in terms of like how people want
to interact with the news and so that's definitely helped um a lot um yeah i think
this is something you can just relate to like that's why i like consumer products and uh not
as much working on like uh business to business products where i like don't understand the use
cases but like news is like oh i look at news i read news and i'm just like that seems wrong or like this seems very
like right this the thing that i think i dislike most that's happening in news currently is this
like editorialization of everything like you're just reading some article on like the front page
of cnn and it's like very editorialized about you know blah blah blah and like you just tell
like these whoever's writing is just like steaming
and wants to drip this
smarmy stuff about stuff and it's like
shut up. What is
the story?
At least we can all
relate to it.
Yep.
Yeah, so I brought up
some people that
whoops.
Hey.
Hey guys, you hear me? Yeah, is I brought up some people that... Hey. Hey guys, you hear me?
Yeah.
Is it Nissan?
Nissan.
Nissan, okay.
Hey, Sarah.
I've actually been following Particle from the Medium kind of blogs that you have been sharing,
along with the social media and whatnot.
Kind of like, I guess, to the beginning days of Particle,
how'd you go about idea validation or
coming to a decision and be like,
yes, Particle is something worth pursuing,
when there's already apps like Apple News,
Google News and stuff like that already in that space.
So kind of like what was your validation framework for the idea?
Yeah, that's a great question.
And in the early days, the first thing we did as soon as we had like something workable is we started a private beta with, you know, friends and families and, you know, investors.
We had raised a seed round early on and, you know, put our investors on it right away
and started to kind of get that very, very early feedback.
The first version of Particle did not actually aggregate stories together across multiple sources.
It was more just like a stream of summaries of individual
articles. And, you know, very quickly we were like, okay, but this is, this is cool because I,
you know, I can kind of quickly get caught up on what a story is about. But what we really want is
the, like, you know, the idea that we can actually take articles about the same topic and bring those
together and summarize it across different sources and therefore get, one, a better summary
because it's, as I mentioned before, it starts to focus on the agreed upon facts between all
those different sources. But two, it starts to give you a map of like,
so you have a story about something
and here are all the contributing publishers
that have contributed to it.
And if you wanna go further, it's now all in one place.
Like, it's a true aggregator in that sense.
And that's, I think one of the things
that really started to click early on with our early beta users is that, like, this is what an aggregator should do.
This is what a news aggregator should do.
It should take all of the news about the same thing, bring it together, summarize it for me, and help me understand if I need to go further or if I'm done and I can move on.
But, you know, we use Apple Test Flight.
We still have a beta going on today
that we, you know, launch our features to early and stuff.
So strongly recommend, you know,
in the early stages of idea creation
to just start to get that feedback as early as possible.
Yeah, gotcha. Yeah, I do have that kind of like the Apple test pilot downloaded as well.
Oh, cool.
Yeah, I'll say aside from that, was there anything like features that people used on the app
that surprised you or, you know, you, you kind of released something and,
and something like really hit it off versus ones that maybe didn't do as well as you thought they
might. Yeah. I mean, I think like we, what we saw the kind of the general usage of the app was like,
you know, people like to scroll through the news. They like to read their summaries. Um,
the news, they like to read their summaries, they like to have it personalized to their
needs and their interests. We added this questions feature so you can ask a question of any story
early on in the beta and we saw that that was really powerful to read a story and decide
that actually there's something else that I need to know about and be
able to get that answer right away. So that was something that we saw the behavior of people doing
that, and that kind of drove us to invest more in that feature. And then during our beta period,
we had expanded the beta a little bit wider. Before launch, we ran a survey and kind of got
information about how people were liking the app
and the different things about the app that they liked. And one of the things that showed up in
that survey that was a surprise to us was that people really liked the quotes feature. So,
basically, in Particle, you can see, you know, the summary of the story, you can see the articles,
where it came from. And you can also see, because we extract kind of like the popular or the important quotes of the story. So if this is like, you know,
a launch from Instagram or something, you might see the quote from Meta or from the Instagram
account that they maybe tweeted or posted on threads or whatever. So we bring that stuff,
we highlight that stuff right away. And I think there's something really nice about seeing the kind of
primary source quotation from, you know, someone who's related to that story, whether that's an
official company account or, you know, Mark Zuckerberg himself or whatever, to like get that
primary source makes you feel more connected to the story or that you're like hearing about it,
you know, straight from the source and elevating that. So making it so that like, again, like you don't have to do any work to find that.
We've taken what we think is most important and we've brought it to the forefront for you.
So it's right there for you.
Really improves the experience.
It makes it like easier to kind of keep up with the news and feel like you're close to it.
So that was interesting.
That was a really interesting insight that we got from our beta.
And so we, you know,
did a bunch of work
to make that feature work better
and make it more efficient
and make a better experience
out of it as well.
Yeah, that's cool.
Because I feel like
the getting data
and like looking at the data
is a lot more useful than like what you think might be good when you when
you make something or when I make something or even when I do something like even when I was
using particle I was like well I'm sick of politics I don't want to like hear about this
I want to hear about science and whatever so I like you know would would swipe on those things
in the the intro part where I was like we'll show you the news that you want. I'm like, these are what I want. And then what I found was
whenever I get a science article, I'm like, I don't know what this is talking about.
It'd be like, like talk about some like esoteric like space thing. I'm like, well,
that's a cool nebula picture, but I don't know what they're talking about. And I actually did
want to hear about the politics because that's kind of what I riff
on X most of the time so I was like
ow. Like I guess I do want
to hear about this annoying stuff
I'm lying to myself
where I'm like I don't want politics I just want
I want to better myself and learn
about technology and science
and it's like I don't really
want to read about that.
I guess I'm too stupid to understand it.
It's interesting to think about what you as a person,
as a human, would like to think you're interested in
versus what you end up reaching for when you have a moment.
And I suffer from that too.
If you ask me or like,
if I sign up for a new app and they ask me like,
what is,
what are you interested in?
I'm not going to pick,
you know,
tabloid celebrity news.
I'm not a person who likes that.
However,
if they start sending me push notifications about tabloid and celebrities,
I am absolutely going to click on them.
you know,
what you say and what you do is not always aligned.
Yeah. And that makes that problem that we were talking about earlier,
about like how the algorithm starts to push you toward a certain direction because you keep
clicking on stuff where it's like, I'm not this person who wants to learn about the stuff. But
I keep watching why Rachel Zegler is a bad actress.
Like, no, no, no, I don't care about that.
But it's just compelling for some reason.
Like, but no, that's not who I am.
So it's like kind of funny.
And we do try to balance that, right?
Because just because you click on something that is clickbaity does not mean that that is healthy for us to just, you know, all of a sudden bombard you with that or take you down that rabbit hole. And, you know, so there is a balance between kind of
those implicit, implicit preferences versus what you've explicitly set. And it matters.
I wonder what kind of data you could like show to the user for that kind of thing. It'd be really
funny. It's like, here's what you think you are. Here's what we can tell from your clicks.
Here's what you think you like. And it'd just can tell from your clicks. Here's what you think you like.
And they'll just be like,
you really, really like, you know,
like this type of content.
You're like, no, that's not me.
Not like that.
But I understand that I click on it a lot.
Yeah.
And, you know, there's a spectrum between like
some of that candy content versus like eating your vegetables
and like, you know,
learning about what's happening in politics today or whatever. And again, like one of Particle's goals is to like,
just bring it down, like it should not feel like eating your vegetables to just keep up with what's
going on. That's true. Yeah. And I think I feel like when Jon Stewart was telling me in the Daily
Show, that's where I got a lot of my news
because that was like enough candy humor
to keep me going to like hear what was going on.
But obviously that tipped me in certain directions
at those times, which is, I don't know.
I still like Jon Stewart.
I still like watching him.
I feel like he's generally pretty fair
compared to a lot of things that you can watch.
But that's just my opinion.
I think you hit it on the nail, right?
Like that was a great entertaining format or whatever.
But it is just one perspective.
And so how can we make it easier?
Because it's work.
It's work for you to go somewhere and get one perspective and then be like, great.
I'm going to go get all that same information again from another perspective just so that I stay balanced.
You're just like, you're not going to do that.
Yeah.
And it needs to be entertaining too, right? Like they don't have jokes.
Like where's the other John Stewart?
They don't have jokes on the right.
Where am I supposed to get my jokes on the right?
Yeah, well, yeah.
That's one thing about AI that I've found
that it doesn't do very well as humor.
It is just like negative funny.
Like the way to make humor
out of ai is to make fun of its attempts at being funny rather than like if you like ever ask it's
like oh make a joke about blah blah or like help me with a post about this or help me reword this
post it comes up with like the most cringe garbage and you're just like oh it's lame yeah oh my god
and like uh i can also like i'm as an influencer i can tell when
uh people are giving me stuff to say that they just were regurgitated from ai they're like hey
like here's a here's a thing you should push and like post this these five different things you
can post and it's like all these like ai things i'm like oh my god like that's cool i'm gonna do
my own thing yeah yeah because you know i'm pretty sure that this was just just like hey
chat you've been taking me five you know using the voice of a crypto influencer it's like
this is the worst that is is super great. Yes.
Nishan, did you have any follow-ups?
Yeah, I guess in terms of... So I've kind of followed Particle on Parallel for any job posting.
So in terms of Particle and AI or even cognition to AI
related to like news consumption? Like,
is there any like specific field or like tech stack in terms of like big data processing
or anything like that, that you could kind of like learn and pick up that would help in this field?
It is like one of the areas that I'm like really interested in. Um, so this is like the perfect opportunity for me to,
uh, ask that. Um, yeah, I mean, we're, we're, uh, uh, we have like a, what I'd, what I'd consider
is like a pretty modern stack. Um, we use Go in our backend. We have an iOS app. We have a website, React, and all that kind of stuff.
We're not really hiring for any specialists right now.
We're a very small company.
Right.
But I definitely appreciate your interest.
If there's anything that comes up, it will show up on that parallel site.
That's cool that you have been following and that you've seen that.
But also, you know, if you want to chat more about that offline, just shoot me a DM and we can connect. No, we should do an interview right now.
Just get on your computer.
We have some questions.
have some questions.
Can you make a hash map of the?
Can you make a hash map?
Yeah.
Yeah.
Yeah.
Yeah, Sarah, I might take you up on that on LinkedIn.
Great.
Yeah, shoot me an up.
Make sure it's.
Oh, and then I brought 51 up because they were requesting.
Do you have a question for Sarah or anything?
No, I'm just kind of listening
and I just kind of was like
trying to get some, you know,
feedback, not feedback,
but more information
on what you guys are, you know,
you know, talking about
more specifically.
But yeah, just came to say hi.
I hope you guys are having a good Friday
and just here to listen
and more education.
So yeah.
Cool.
Awesome.
Thanks for joining.
Yeah, thanks for joining 51.
We're almost at the hour.
So we will, I guess, summarize as a help for 51 to get good information.
So I added the Particle News link to the App Store app.
As far as I know, it's free.
And you can just use it and it's uh it's pretty
cool um it is free yeah it's pretty downloaded from the app store pretty easy to use um one of
the first things that makes you do is like tinder swiping on news articles which is fun
it's like i don't want this this is crap and then it's like yeah i want this um i don't know the tinder swipe mechanism it was genius even though it destroyed dating and ruined ruined 20 somethings
forever but the the concept of it and uh how easy it is to do is is no matter what the app is i'm
always like this is fun um yeah we've got a couple other uh folks requesting. Do we have time to just take a couple more?
Yeah, I'll just bring everyone up.
Yeah, why not?
Bring a baby.
How's it going?
What's up? What's up?
How's it going?
Yeah, it's been a great space.
I hope your hand is feeling better, by the way.
That's rough.
I was able to fold my clothes.
I'm more confident with my left hand, I guess.
There you go.
Did you have a question for our panel of Sarah?
Yeah.
I mean, insightful.
I think it's valuable to get all sources of news um especially now in
this day and age so it's really cool i'm definitely going to check it out i had two questions one is
um just when you were going and getting investors like how did that go about did you know people
and you reached out did you have different pitch decks just curious on how that process went and
then um secondly if you guys have ever
used Claude Code within the terminal, because I've been using that to build an app and it's been
really... Am I clapping? Is that me? Wow. That was Doge C. Please don't clap. This is the opposite
of Jed Bush. Don't clap. Yeah, I was just curious if you've ever used Cloud Code within the terminal.
I've been building or trying to build out different AI apps using it,
and it's been way better than Cursor from what I've used it for so far.
Yeah, so I'll address your first question first.
So when we did our seed fundraising, um, you know, we did have a prototype and
a deck and I think, uh, having something that, especially if you're working in consumer,
um, having something that people can hold and play with and feel, you know, goes a really,
really long way.
Um, so I think that helped.
Um, and then we had, you know, connections, um, from to, to VCs just that we've built over, you know, the, you know, 15 years
or whatever that I've lived in Silicon Valley, you know, just kept up relationships with VCs.
And so we started with those. So, you know, I could talk a lot more about that. So maybe I should at some point.
With respect to Cloud Code, I actually haven't tried Cloud Code yet.
I have heard great things about it.
And I believe you.
I believe you that it, you know, for at least some set of use cases or maybe even many use cases, it might work better than Cursor.
So that's great.
I'm glad to hear you're having success with that.
I can't keep up with all of my paid cursors so i gave him a year it's worth of money so i gotta go with that i
don't want to keep buying stuff and being like yeah this one's better this and then like you
know a month from now like cloud code will suck and then curse would be good again and then another
month after that like cursor will suck again. It's just seems like this is going.
Cursor, like, I don't care if it sucks.
I don't care if you ever go on like Reddit's cursor thing.
Just people are just complaining all the time.
There's nothing good about cursor is the worst thing ever.
So I read that.
I'm like, oh, maybe I should use something else and get insecure.
But then I use it.
I'm like, it seems fine.
But yeah, it just depends on what
you're trying to do i think yeah cool thanks thanks for the question do you have any uh follow-ups
before we go to one of the doge coins questions yeah no no follow-ups uh thanks for the insight
overall yeah it's nice having a you know actual uh founder up here to talk about stuff that they know about compared to when I talk about like, oh, this might happen.
I don't know.
I don't know how anything works in the real world.
But you are a cultural tastemaker, Billy.
So what you say actually probably becomes like, you know, you like will it into existence.
Yeah, I really don't like, though, when I make like joke posts and then the news picks it up some way
or whatever i'm just like oh my god and then it's like all super clickbait too it's like
here's what the creator of dogecoin said about the price of bitcoin and all this like long
article and then like in the middle like what i said was like bitcoin price goes up and down who
knows and like who reads this Who's reading this garbage?
You wield great power on your X account.
With great power comes great responsibility.
It's the worst.
Dogecoin
Algerian
did you have any questions
for us?
Or is this a
thing of support and hearts?
It might just be...
Assalamu alaikum.
Now English, I am speaking Arab, Habibi.
I don't speak Arabic, actually.
But Assalamu alaikum to you.
And thanks for joining.
Yeah, thank you for joining us.
Thanks, Sarah.
Last question.
Sure.
Hey, so what's been like the most unexpected role
that you've had to play as like a ceo uh the most unexpected role um um yeah earlier on when
i was getting office space i think just dealing with like dealing with like the logistics of like getting an office, uh, was,
yeah, something that I had not had to deal with before. Like, like we were subleasing from like
another company, a crypto company actually. And, um, uh, you know, they like wouldn't get back to
us. And then so we'd go back to like the building management company and like, yeah, just, just
dealing with that real estate is commercial
real estate is the slowest moving thing i think that's that's good to know yeah i think uh i
remember the only time i ever looked at commercial real estate was when i was like uh maybe i can
live in one of these places and then i I found that that was illegal. Like,
but what if,
but what,
cause it was like cheaper.
It was like a lot cheaper than residential real estate in Bay area and residential real estate in Bay area is obscene,
but just have to,
you know,
fake,
make a company and then fake live in my office.
And I don't know how I would wash myself.
Yeah.
Yeah. Very half-baked idea.
Cool. I guess if there's nothing else,
I guess we can call it for the day, but thanks everybody for joining the spaces.
Thank you so much, Sarah, for sharing as much as you shared. And yeah, everyone check out Particle News,
especially if you have an iPhone, you can just get the app. It's pretty neat.
I love the branding, by the way.
Oh, great. Glad you like that. Yeah, we've been trying to be thoughtful about how we brand it.
So that's awesome.
I like the branding of your profile.
That's a cute Doge.
There are a lot of Doges, and some are more cute than others.
And that's a really cute one.
Yeah.
What about mine?
Which one's better?
I mean, yours is iconic.
Yours is like you can spot Billy from a mile away.
This one looks like maybe it's like AI generated or something.
It looks fun.
Yeah, that's the best part about AI is it's allowed me to fake draw things
since I have no art ability whatsoever.
But I can tell a prompt to do something and have it look generally good
and then have people be like, wow, it ai slop and then they get mad yeah my profile picture right now is actually ai from one of those
like headshot uh but i think that's good like because you know beats like paying someone you
know a thousand dollars for a bunch of headshots to take and then i think the worst part about that
is just like having to pose for headshots.
It's just like the worst.
But then you just take a picture of yourself
and be like, hey, put me and put my background
in this in the Maldives or something
and make me look great.
And you can do that.
So thank you, AI, for helping us with that
and for helping us with news.
And yeah, thanks for having me.
Thanks for coming to the space.
Yeah, of course. yeah all right bye everybody bye
thank you Thank you.
