How we built a 1000x Growth Product in 1 Year | Perplexity AI, Aravind Srinivas

I'm Arvind Srinivas.

I'm the co-founder and CEO of perplexity AI. Perplexity is a conversational and

search engine that aims to deliver answers to you, to whatever questions you may ask.

We are trying to revolutionize how people consume information online.

Instead of getting ten blue links, they can just ask questions in natural language

and just get it answered instantly.

And we launch the product on December 7th, 2022.

We have like about 10 million monthly active users at this point.

It's basically grown thousand X over a period of one year.

So I grew up in India, studied in one of the IIT's there,

and I was really into algorithms programing ever since the beginning.

A friend of mine told me about a machine learning contest, which I didn't even know

what machine learning was, what?

All they told me was, hey, there's this data set and you can figure out a way

to predict the output given the input.

And it was fun.

And I won the contest and I didn't spend a lot of time on it,

and it came more naturally.

So I decided to go deeper into it.

And I went and did my PhD in Berkeley on AI and deep learning.

I worked at OpenAI in 2018 summer as a research intern.

I thought I was good, okay, I did really well in India.

I came to Berkeley.

I'm like, definitely one of the top AI PhD students.

And then I went to OpenAI and I felt like really bad because people

were so much better than me.

It was a big reality check that, okay, I could improve a lot more in programing.

I could improve a lot more in first principles.

Thinking my clarity of thoughts.

After an internship at OpenAI in 2018, that was when GPT 1 was published.

We realized that there is this new form of learning using all the internet data

and learning from it, and I figured that was going to be more important.

So I told my advisor that this is the right thing to do.

We should go work on this.

And he was actually like pretty open minded and said, okay, you know what?

Like I'm not a specialist here, but let's try.

I mean, if this is the next thing, the best way to learn a new topic is

to force yourself to teach it to others.

So we spent a lot of time holidays, weekends, just learning and coding

and just understanding all these things.

And we did this for two years.

All that helped me find a new research topic, which is how to combine

generative AI and RL together, which is what results in these amazing

technologies like ChatGPT.

ChatGPT is not just predicting the next word on the internet.

It's doing that and then making sure that you know how to communicate with humans.

I'd always been interested in entrepreneurship

because I've been in the Bay area.

I watched this TV show, Silicon Valley, which is pretty real, but never really

found an example of an academic turned entrepreneur that I really resonated with.

It was all like undergrad dropouts.

At one point, I was in the library in the late nights reading books,

and then I stumbled upon this book, wrote the story of Larry and Sergey

in the book How Google Works.

Larry had written the foreword.

In it, I had only two career pathways for myself.

It was either to be a professor or an entrepreneur, and the reason is

that no other career pathway would let me execute on my own vision.

I would have to be working on someone else's vision.

I wouldn't be able to bring out what the ideas I have in my head into reality.

Artificial intelligence would be the ultimate version of Google.

So we had the ultimate search engine.

It would understand everything on the web, it would understand

exactly what you wanted, and it would give you the right thing.

Yeah.

Perplexity is the world's first conversational answer engine.

What does that mean?

Earlier, we were used to entering something like keywords

or a bunch of phrases, and Google gives you ten blue links and

you open each of them and start reading.

Perplexity is trying to build a future where you don't have to do this.

You can just come and ask a question, just like how you would ask a friend,

and that AI replies to you with the answer, but not just the answer.

Every sentence that it says also has a corresponding reference,

or we call it a citation.

This is all coming from our academic background.

Like my co-founder, Dennis and I are PhDs.

We figured that we would use this principle that everything in a paper that

you write in academia, you have to back it up with reference from some other paper.

And that's how perplexity works.

It's almost like how a journalist essay is written or research paper is written.

Often you're curious about something, but you don't exactly know what you want.

Even so, how can I help you if you don't know what you want?

People are not expert, prompt engineers they're never going to be.

Don't blame the user for not having a good prompt.

Blame the AI for not being able to expand or help them expand themselves

to a good prompt.

That's why we built this thing called copilot on our side, where as you

ask a question, copilot will last.

Clarifying questions on your prompt is basically getting expanded interactively.

This is similar to talking to a friend like, hey, you know what?

I'm figuring out which school to go to.

I was like, oh, okay, cool. What are you actually interested in?

Are you interested in like, English majors?

So you're interested in computer science?

And then I think I might be interested in both English and computer science.

Okay. Yeah.

You know what? Yale might be a good option for you.

Like, that's how you talk to a friend, right?

We want that experience to come to a search engine,

to that human intelligence needed to do that is being done by an AI now.

And we think this is the future of how people are going to interact

with information on the internet.

We launched the product on December 7th, 2022, our first day,

I think we saw around 2000 3000 queries.

Now we serve more than 3 to 4 million queries a day.

It's basically grown thousand X over a period of one year.

Growth so far has been that somebody says ChatGPT doesn't work for this particular

thing, or like Bard sucks at this thing.

And then like, people just tweet, oh, look at this perplexity thing.

It just gets it. Look at this thing.

This is how we maintain the quality of the answer comes down to improving every

single component here a component of like, does it have spammy sites

or does it have like high quality sites?

How good are you at writing that amazing, concise summary without hallucination?

We are playing the orchestra here.

These are all like individual musicians and any one musician failing

will make the result fail.

That's why this is a hard thing to build.

That's why this is not something where, oh, because you're a startup,

you're going to lose.

Because even for a big company, playing the orchestra is hard.

Of course, if you have more money, you can hire better musicians and like you

play a better orchestra over time.

But that's still the part of orchestrating.

But the user doesn't care where it goes wrong in any of these.

For the user, they read an answer and they're like, oh, this is good.

Or like, this is not good, right?

So that's why this particular product is super hard to build.

And that's why, like, we are so focused on improving every aspect of this.

This is a really hard problem.

And we believe it can be solved over time as we gather more data from users

as improve our own like stacks in each of these components, your experience is going

to keep getting better.

So the Pro plan is priced at $20 a month.

It's the exact same pricing as ChatGPT plus, I'll tell you why.

So we use OpenAI's GPT four.

If we priced it lower than chat GPT plus, people would come and pay for it,

but not necessarily for what we're probably because we subsidized GPT 4

and gave it to the user.

And subsidy in any industry has product market fit.

But then do you have product market fit as a company because you're subsidizing

something everybody wants, which is GPT 4, or do you have product

market fit for your core offering, which is combining LLM and search together?

And it's very important for you to not conflate something with something else.

So we decided, okay, we'll price it at the same price and then see how many people

are still paying for our product, because they realize that we are the best provider

of search and algorithms together.

Either they have to cancel GPT subscription, come here,

or they have to pay for both.

Just like how you pay for both Netflix and HBO.

That's why we decided to do this, and we are super happy that that worked, because

that means what if a user comes and pays for us, communicates to us one thing,

which is they value that you are providing the best service of this one particular

thing, that they want the highest quality.

Yeah, the best strategy for startups is to focus on very few things,

like literally even one thing, because there's not much time.

As a startup, you're supposed to move fast, and as a startup you have

very few shots at failure.

You're also supposed to ship high quality things so that the user trusts you.

So physically impossible for you to do many things.

We're still a small team, around 30 people.

When you have fewer people, you can only do fewer things.

So therefore you spend a lot of time thinking about what to do.

And once you've decided, you just do it.

There is a quote I really like from the Airbnb founder

that you have to earn the right to ship a new feature from your user, because

the user wants already a bunch of things.

Your job is to actually go and do that for them, and once they are happy,

you're like, hey, give me new features when you're doing pretty well.

That's when you got to go and ship new features.

We have this mentality in the company that don't immediately say yes

to every single obvious idea that you can do, try to really think about

what the user wants and how does it work in the context of our mission, which is

to make the world's most knowledge centric company ultimate knowledge app

And once we have strategized that well, we would just focus on execution.

We wouldn't get distracted.

And once it's shipped and it's in a good enough state, we would finish the project

and move on to the next thing.

And then it becomes like a repeatable workflow.

And it's also the culture you want to set.

Like you tell people, okay, I'll do it tomorrow.

Why can't you do it today? Just ask that question.

Don't tell them to do no, no, no, you do it today

because just ask can we do it today?

And if they have a solid explanation for why it cannot be done today, then fair.

But maybe they didn't even consider they thought okay, they could do it tomorrow.

So not in a way where it comes across as toxic, but more

like trying to push them towards urgency.

Hey look, we are a startup. We need to execute.

Like if we don't, all our potential just decays.

If you have a rolling ball and you do nothing, it will automatically stop.

But if you have a rolling ball and you keep kicking it, it'll go even faster.

Something is complex because there's a lot of information.

Then force your brain to say, okay, this is a lot,

but what is the one most important thing?

What is the second most important thing? Usually there's not more than two.

Let's say there's like one thing that has two choices.

And there are like three things that are eight choices.

Now your brain is not able to process eight choices at once.

It's usually has 3 or 4 at best.

So your job is actually to figure out what is that two choices.

In fact, there is like a advice from Reid Hoffman that says, in life,

whenever you're going to make decisions, people usually do pros and cons

where they write down the pros, they write down the cons and then see

which has more, and they pick that option.

But that's like the wrong way of doing things, because that way you're weighing

everything equally important, where things are not equally important.

Usually some things are way more important than others, so you've got to be able to

take something and pick the most important thing out of it and focus on that.

Usually it works.

Reformulate the problem better, and so the complex problem becomes

much simpler and then iterate.

Look, I'm not saying I'm really good at this today.

I can still improve and so can everybody. So believe in the improvement process.

Don't believe in like being perfect. And we all learn.

We all make mistakes and it's fine.

So I've given this advice in other interviews I want to continue

to say this not just for consistency.

I really believe in it.

When you are starting a company, do what you really love because the world

is not something that's static, it changes dynamically really fast.

I would say what you love doesn't usually change, so start with that.

The mission is not about making money.

That said, the mission requires money and therefore we will make money

in order to like serve the mission.

The metric should never be like oh, by X year or X month.

I'm going to increase the valuation by alpha times X. It should be really focused

on okay, I should make the product better.

I should have more users.

I should have a higher quality product, more accuracy.

A lot of people, when they wake up, they feel like going back to bed.

They feel like want to sleep 1 or 2 more hours more,

and nothing's really going to change.

For me, it's the opposite.

I'm like waking up sooner than I wanted to, sleeping later than I wanted to.

When the day ends, I always feel like there's more stuff I could have done,

so that's actually a privilege.

I also feel stress, but the opposite wouldn't make me feel

any fulfillment, honestly.

So it's very fulfilling.

It's definitely a privilege and I want to keep going this way.