전사 🔗
WEBVTT
00:00:00.160 --> 00:00:10.880
Instead of like you having to actually you know run a very long and extensive process and finding people that are not even a fit, JuiceBox ensures that you have a good top of funnel of people that are actually a good fit.
00:00:11.039 --> 00:00:13.279
If we make sure that you're engaging the right people.
00:00:20.320 --> 00:00:25.920
Welcome to InProd, conversations with founders and operators who are building the next wave of AI companies.
00:00:26.160 --> 00:00:31.039
We go deep on product execution, go-to-market, and the messy middle of building a business.
00:00:31.280 --> 00:00:38.399
If you're an aspiring founder, operator, or just startup curious, this is your inside look at how companies actually get built.
00:00:41.119 --> 00:00:46.240
Before we get to the episode, Laura and I are excited to announce a partnership with Overlap.
00:00:46.640 --> 00:00:56.240
Fresh out of Y Combinator's Summer Batch, Overlap is an AI-driven app that uses large language models to curate the best moments from podcast episodes.
00:00:56.479 --> 00:01:08.480
Imagine having a smart assistant who reads through every podcast transcript, finds the best parts or parts most relevant to your search, and strings them together to form a new curated stream of content.
00:01:08.719 --> 00:01:10.719
That is what Overlap does.
00:01:11.040 --> 00:01:18.959
I personally use Overlap to explore content on future podcast guests and highly recommend it as another diligence tool for investors.
00:01:19.280 --> 00:01:21.680
Check out Overlap on the App Store today.
00:01:22.079 --> 00:01:25.359
Today we are joined by Ishan Gupta, co-founder of JuiceBox.
00:01:25.519 --> 00:01:31.519
Juicebox is an AI-powered recruiting platform that recently raised its 30 million Series A led by Sequoia.
00:01:31.760 --> 00:01:32.480
Welcome, Ishan.
00:01:33.120 --> 00:01:34.000
Very glad to be here.
00:01:34.079 --> 00:01:35.200
Thank you so much for having me.
00:01:35.439 --> 00:01:36.400
Yeah, thanks for coming.
00:01:36.719 --> 00:01:38.480
So let's go back to the early days.
00:01:38.719 --> 00:01:42.879
When did you first become interested in tech and how did that shape your high school experience?
00:01:43.280 --> 00:01:47.040
I first became interested in tech around middle school, I would say.
00:01:47.200 --> 00:01:49.280
The first thing I got into was comparative programming.
00:01:49.439 --> 00:01:55.359
So I I found this concept pretty interesting that there's people who compete on writing code and solving problems more quickly.
00:01:55.519 --> 00:01:57.280
And I always enjoyed solving problems.
00:01:57.439 --> 00:02:00.400
I was always into math, computer science in in school as well.
00:02:00.560 --> 00:02:03.040
So it's like I first got into comparative coding.
00:02:03.200 --> 00:02:06.799
So there was a lot of my middle school and some of my high school.
00:02:07.040 --> 00:02:11.039
And then at some point, I really got interested in building applications.
00:02:11.199 --> 00:02:14.800
So it's like maybe I should try building web applications, mobile applications.
00:02:14.960 --> 00:02:18.000
And I used to used to game a ton as well on my phone.
00:02:18.080 --> 00:02:21.599
So I was like, what if I can build more mobile games and more mobile applications?
00:02:21.919 --> 00:02:25.199
Started doing that around the beginning of high school.
00:02:25.439 --> 00:02:30.960
And at some point, I was like, okay, what if I can I just want to see if I can make some money out of it.
00:02:31.039 --> 00:02:35.599
So I was like, oh, I'll start a freelancing thing and let's see if I can do this for other people.
00:02:35.840 --> 00:02:37.199
And that actually worked out very well.
00:02:37.280 --> 00:02:46.960
So I managed to scale a pretty good freelancing business of just finding these clients online, going to them, talking to them, asking them if they need a website built or an application built, and then just doing it for them.
00:02:47.120 --> 00:02:48.800
And this was all around COVID.
00:02:48.960 --> 00:02:50.159
So like school was remote.
00:02:50.240 --> 00:02:53.360
So then I was like, okay, I don't really have to go to school anymore.
00:02:53.439 --> 00:02:54.400
So what do I do at home?
00:02:54.479 --> 00:02:56.319
So I just started doing the whole freelancing thing.
00:02:56.479 --> 00:02:58.080
That's initially how I got into tech.
00:02:58.400 --> 00:03:01.439
Okay, well, that makes a lot of sense given what you're building today.
00:03:02.159 --> 00:03:06.319
But before we get into that, is that initially how you and your co-founder met?
00:03:06.879 --> 00:03:08.800
That's also actually a very interesting story.
00:03:08.960 --> 00:03:16.560
So I grew up in India in this town called Kanpur, and my co-founder was starting at Harvard at the time here in the US.
00:03:16.719 --> 00:03:17.759
This is during COVID.
00:03:18.240 --> 00:03:21.840
He organized a competition called the Crown Education Challenge.
00:03:22.080 --> 00:03:26.159
I came across this thing online, and I kind of knew at the time that I want to go to the US.
00:03:26.240 --> 00:03:29.199
So I was looking into competitions and different things people do here.
00:03:29.360 --> 00:03:35.840
And I came across this thing where you have to submit a research abstract, like some sort of a technical solution to something pandemic related.
00:03:36.000 --> 00:03:41.439
The thing that stood out to me is that the prize was you get to meet Stephen Wolfram, who is a very popular computer scientist.
00:03:41.520 --> 00:03:43.039
He's the founder of Wolfram Alpha.
00:03:43.199 --> 00:03:44.719
And I was like, oh, I want to meet this guy.
00:03:44.960 --> 00:03:45.919
He's pretty amazing.
00:03:46.080 --> 00:03:47.360
So I took part in the competition.
00:03:47.439 --> 00:03:48.159
I ended up winning.
00:03:48.240 --> 00:03:49.680
So I went and met Wolfram.
00:03:50.080 --> 00:03:55.599
But the more fun thing that
Load full transcript
came out of it is I met David, who was one of the organizers of this competition.
00:03:55.840 --> 00:04:00.800
David and I started talking and we realized that both of us share a lot of interests.
00:04:00.960 --> 00:04:08.479
He was working on the growth team at Snapchat at the time, and he had also kind of realized that he doesn't really want to work at a very big company and he wants to do something of his own.
00:04:08.639 --> 00:04:10.319
And I really enjoy building things.
00:04:10.479 --> 00:04:19.519
So we got in touch, we started working on a bunch of projects together, and then eventually at some point when we had the confidence that we really worked well together, we decided to like build a company.
00:04:19.839 --> 00:04:20.879
Very cool story.
00:04:21.120 --> 00:04:25.439
Maybe before we dive into what you guys are building, tell us a little bit about the name.
00:04:25.680 --> 00:04:31.680
Juice Box is obviously a fun name that you can remember, but I'm curious, is there some other backstory behind it?
00:04:32.079 --> 00:04:34.079
Yeah, so it's it's actually funny.
00:04:34.240 --> 00:04:38.240
The first thing my co-founder and I built together was a music app.
00:04:38.480 --> 00:04:41.279
And JuiceBox was a wordplay on Jukebox.
00:04:41.680 --> 00:04:52.879
And it was it was like this like TikTok meet Spotify experience where you can discover music and it like learns from your taste and you find like more music that you or artists that you'll you tend to like based on your existing taste.
00:04:53.120 --> 00:04:55.920
That's what we started off with, and it was called Juice Box.
00:04:56.160 --> 00:05:05.040
Then we had a long pivoting journey when we actually started the company, and at some point we were about to launch our recruiting product, and we're like, okay, no one's gonna pay attention.
00:05:05.120 --> 00:05:07.439
It doesn't matter what we call it, let's just stick with juice box.
00:05:07.680 --> 00:05:09.680
We have the domain, anyways, let's just do that.
00:05:09.759 --> 00:05:11.279
So we called it juice box.
00:05:11.439 --> 00:05:16.560
And then I don't know why, but we got a lot of message fit instantly, a lot of traction.
00:05:16.720 --> 00:05:21.519
And people started recognizing the name Juice Box, which is why we ended up sticking with it.
00:05:21.680 --> 00:05:25.040
At this point, I think it's people enjoy like the fun aspect of it.
00:05:25.120 --> 00:05:29.680
I think a lot of HR and recruiting is extremely boring and like very corporate-coded.
00:05:29.759 --> 00:05:33.519
So I think people enjoy the fact that there's something more fun about us.
00:05:34.079 --> 00:05:40.720
Yeah, well, it's definitely a playful name, which I think is really cool in like such a, like you said, HR kind of talent recruiting space.
00:05:40.959 --> 00:05:48.000
You mentioned obviously you and David meet through this competition, you decide, hey, we want to build something, but it sounds like you pivoted a bit.
00:05:48.160 --> 00:05:55.040
Maybe talk us through kind of that pivot story and also curious where YC plays in this journey as well.
00:05:55.279 --> 00:06:00.399
So YC was incredibly helpful in the first part of our journey because we started off with consumer.
00:06:00.480 --> 00:06:02.240
Like we were doing the music app, right?
00:06:02.399 --> 00:06:08.800
And that's what most young founders, I think, start off with, especially anyone who like drops out of school, because that's the first thing you interact with, right?
00:06:08.879 --> 00:06:11.759
Like you use a lot of consumer applications and then you're like, oh, I can do this better.
00:06:11.839 --> 00:06:13.360
So I'll build a consumer app.
00:06:13.600 --> 00:06:14.959
That's what we started with.
00:06:15.199 --> 00:06:28.240
YC was very helpful for us to realize that there's this whole world of business software where it's not just about distribution and doing things at scale, it's about really deep diving into someone's problem and building a novel solution for it.
00:06:28.480 --> 00:06:31.680
So I think that is what YC was incredibly helpful for us with.
00:06:31.839 --> 00:06:35.439
The first thing we started off in YC was actually a recruiting marketplace.
00:06:35.600 --> 00:06:38.240
Because as I talked about, I freelanced a lot in high school.
00:06:38.319 --> 00:06:44.079
And my co-founder David also did a lot of jobs with different companies, cut jobs while he was in college.
00:06:44.240 --> 00:06:47.920
We ended up realizing that we really had to go out of our way to find these jobs.
00:06:48.079 --> 00:06:50.639
Like it's not like it was very readily available.
00:06:50.800 --> 00:06:57.199
So our initial intuition was that we should build a freelancing marketplace that helps connect people for these roles more easily.
00:06:57.360 --> 00:07:04.720
We started off with that doing YC, and then quickly ended up realizing that there are a lot of marketplaces that do roughly the same thing.
00:07:04.800 --> 00:07:10.399
So there was there's no real way for us to build a 10x better product, which is why we ended up pivoting.
00:07:10.560 --> 00:07:16.560
But the good thing that came out of us building the marketplace is that we really understood the recruiter persona.
00:07:16.720 --> 00:07:27.600
We worked with a lot of talent teams, and we quickly realized the main problem, or like the most high value function in recruiting, is finding the right people and engaging them, getting them into your process.
00:07:27.759 --> 00:07:35.439
And we realized that the recruiters are spending a lot of time on this, which is why we ended up realizing that that's what we should be building a solution to.
00:07:35.759 --> 00:07:43.839
This is around the time ChatGPT also came out and LLMs really started taking off, and GPT 3.5 and then four, and things started getting better.
00:07:43.920 --> 00:07:50.879
And we realized that these LLMs can help with a lot of this top of funnel process
in recruiting, which got us to the product that we are at today.
00:07:51.199 --> 00:07:54.800
And I think that's a good transition into sharing what the product is today.
00:07:55.199 --> 00:07:58.800
What does it look like and how are folks using it just to ground the audience?
00:07:59.040 --> 00:07:59.199
Yeah.
00:07:59.360 --> 00:08:04.639
Juicebox is an AI recruiting platform that helps companies find and engage talent using LLMs.
00:08:04.879 --> 00:08:20.879
So the traditional, what the traditional recruiting flow looks like is you align with some requirements with your hiring manager, you go into a traditional platform, you will put in a bunch of filters, you will start looking at profiles, and then you assess them one by one, you shortlist the people you like, and then you send them an email.
00:08:21.120 --> 00:08:26.319
And that process takes hours because you have to look at every single profile one by one.
00:08:26.560 --> 00:08:28.720
And it's an extremely time-taking process.
00:08:28.879 --> 00:08:32.159
With JuiceBox, we try to automate that using LLMs.
00:08:32.320 --> 00:08:45.360
So instead of you having to do the whole searching process and going through profiles one by one, you describe in natural language who you're looking for, and then we surface the best candidates for you because LLMs are able to evaluate those profiles and surface the best ones for you.
00:08:45.600 --> 00:08:49.840
Beyond that, we build recruiting agents that take even more control over that process.
00:08:50.000 --> 00:08:54.799
So instead of you even having to run that search, you kind of just calibrate an agent like a recruiter.
00:08:55.039 --> 00:08:56.320
They would go back and forth with you.
00:08:56.480 --> 00:08:58.480
The agent would try to understand your requirements.
00:08:58.559 --> 00:09:06.720
And once it's like fully calibrated, you can just let it run the process, reach out to people on your behalf, and try to get them into your recruiting funnel.
00:09:07.440 --> 00:09:08.639
That's very exciting.
00:09:08.960 --> 00:09:16.320
How large is your database today and where does it span to in terms of an industry, a geographic standpoint?
00:09:17.039 --> 00:09:19.120
We have over 800 million profiles.
00:09:19.360 --> 00:09:23.519
We collect a lot of like professional data across the web from a lot of different sources.
00:09:23.679 --> 00:09:32.480
So it spans a ton of different geographies, pretty much every single person who has a professional presence on the internet, we likely have them in our database.
00:09:32.799 --> 00:09:38.399
And are you finding you mentioned the flow that a recruiter, one of your customers would take to utilize JuiceBox.
00:09:38.879 --> 00:09:48.879
But as you guys were kind of building the company in the early days, could you talk us through maybe an aha moment, a feature that you launch where people immediately you realize, hey, we maybe have PMF here.
00:09:49.039 --> 00:09:50.559
Yeah, just curious how you experimented.
00:09:51.039 --> 00:09:55.279
The feature that people really liked was this feature we have called Autopilot.
00:09:55.360 --> 00:10:00.960
And any recruiter who uses JuiceBox is like always raving about it because they they really enjoy using that piece.
00:10:01.120 --> 00:10:14.000
And the best the autopilot feature we have is that that was the first thing we launched where instead of like actually having to like just summarize profiles or review them, you actually are able to stack rank them based on different criteria.
00:10:14.240 --> 00:10:22.720
So when a hiring manager specifies their criteria or a recruiter does a search, it's a lot more complicated than it would seem to someone who doesn't work in talent.
00:10:22.799 --> 00:10:29.919
Because there's there are like must-haves, there are uh nice to have's, there are like bonus points you would maybe like in some profiles.
00:10:30.080 --> 00:10:33.039
And sometimes it's even a process to like find these requirements.
00:10:33.279 --> 00:10:39.440
The thing that the autopilot does very well is it uses LLMs to like semantically look at every single one of these profiles.
00:10:39.519 --> 00:10:41.519
So you're not just going off of keywords anymore.
00:10:41.600 --> 00:10:43.919
It's not like I want to find people at Google, you know.
00:10:44.080 --> 00:10:47.840
It's more semantic than that, that you can type in more nuanced searches.
00:10:48.000 --> 00:10:48.720
That is not possible.
00:10:48.879 --> 00:10:53.440
So I think that is the feature that really stood out to a row of recruiters and they really enjoy it.
00:10:53.679 --> 00:11:04.559
And the PMF point for us when we like kind of realized that we had PMF was when I remember there was like a month where we had to focus a lot of our energy on fundraising and some like product-related things that were not working.
00:11:04.720 --> 00:11:12.879
And at the time, I think David, my co-founder, was pretty much our only seller, and revenue growth was still pretty strong, even though we didn't put a lot of effort into sales.
00:11:12.960 --> 00:11:18.399
So we started realizing that it's it's organically sitting well with people, and all of our growth is coming from word of mouth.
00:11:18.480 --> 00:11:20.559
It's literally people telling each other about it.
00:11:20.720 --> 00:11:24.720
So I think that was the moment we realized that okay, we actually have something pretty good here.
00:11:24.960 --> 00:11:37.759
It's impressive because I think a lot of the best companies we work with benefit a lot from, like you said, word of mouth, that viral loop that happened, and HR and recruiting professionals in particular, at least my understanding is they're they're a close-knit community.
00:11:37.919 --> 00:11:41.919
So it's great that you're already starting to see them talking about the platform.
00:11:42.080 --> 00:11:43.440
Maybe just one more question on that.
00:11:43.600 --> 00:11:47.279
From a customer
perspective, who are the types of folks that utilize Juice Box?
00:11:47.360 --> 00:11:53.200
Is there a particular type of customer that you're like, hey, if you're listening to this now, you should be using JuiceBox today?
00:11:53.440 --> 00:11:53.919
Yeah.
00:11:54.240 --> 00:11:57.440
Every company hires, and we have customers all across the board.
00:11:57.519 --> 00:12:01.840
We have Fortune 500 companies where their entire talent teams rely on JuiceBox for sourcing.
00:12:02.000 --> 00:12:12.720
We have a lot of like high-growth companies like Ramp is a customer of ours, Varcata, ScaleAI, a lot of these customers, high growth tech teams that really need to find top talent.
00:12:12.799 --> 00:12:14.399
They utilize juice box to do that.
00:12:14.559 --> 00:12:16.320
We have over 3,000 customers.
00:12:16.480 --> 00:12:21.519
So anyone who feels like recruiting is a need for them, they should absolutely be using JuiceBox.
00:12:21.759 --> 00:12:29.200
Because we have product-led growth, there's also a very strong self-serve motion, which is why we don't really need to spend that much effort on outbound.
00:12:29.279 --> 00:12:34.559
In fact, we almost do very little outbound, which is why a lot of our growth is pretty much inbound.
00:12:34.639 --> 00:12:37.360
So that's literally customers telling each other about it.
00:12:37.440 --> 00:12:40.080
And as you mentioned, these communities are very close-knit.
00:12:40.240 --> 00:12:46.080
So recruiters like to talk about their stack, they like to share it with other recruiters, which has worked very well for us.
00:12:47.440 --> 00:12:49.919
Now, a quick word from our sponsor.
00:12:50.240 --> 00:12:56.559
Laura and I are thrilled to announce a partnership with Grata, an intuitive deal sourcing intelligence platform.
00:12:56.879 --> 00:13:00.879
Keep missing out on private market deals, ditch the Rolodex and get Grata.
00:13:01.360 --> 00:13:09.519
Grata's AI-powered platform provides accurate data, deep insights, and time-saving workflows that help get more deals done.
00:13:09.759 --> 00:13:14.960
And it's trusted by over 500 firms in 3,000 deal makers, including myself.
00:13:15.279 --> 00:13:24.879
I've personally used Grata in both my deal sourcing and diligence workflows and would recommend their tool to any investor across private equity, growth equity, corpdev, or investment banking.
00:13:25.120 --> 00:13:28.000
We've negotiated a special offer for our audience.
00:13:37.759 --> 00:13:39.519
Now back to the episode.
00:13:39.919 --> 00:13:42.960
How do you encourage stickiness within your customer base?
00:13:43.120 --> 00:13:53.600
Because I'm sure you're working with larger Fortune 500 customers where it's like, okay, we always have an open engineer role or an open marketing role, and so it makes sense where if it's working, you're not going to replace what's working.
00:13:53.759 --> 00:14:01.039
But maybe for smaller companies where the hiring is a bit more sporadic, how do you ensure that folks are coming back to juice box?
00:14:01.120 --> 00:14:07.360
Because I'm not sure if all 3,000 are paying, but I'm sure that you're trying to upsell and encourage as many as possible too.
00:14:07.440 --> 00:14:07.600
Trevor Burrus, Jr.
00:14:08.080 --> 00:14:10.480
All 3,000 of them are paying customers, typically.
00:14:17.440 --> 00:14:17.679
No.
00:14:18.000 --> 00:14:22.159
I think the there's two elements to the stickiness base that you're talking about.
00:14:22.320 --> 00:14:28.159
Even though yeah, for the Fortune 500s, they have so much attrition that they need thousands of roles filled every year.
00:14:28.399 --> 00:14:30.639
So it's it's a whole different scale of problem.
00:14:30.879 --> 00:14:34.080
For a lot of the smaller companies, it is a lot more cyclical.
00:14:34.159 --> 00:14:37.039
You're hiring more sometime, you're hiring less another time.
00:14:37.200 --> 00:14:41.440
There's two things that make the product incredibly sticky, even in those cases.
00:14:41.679 --> 00:14:42.879
One is the data.
00:14:43.039 --> 00:14:50.080
So you have all of this data within juice box about who you've engaged in the past, who responded, who's interested, who's not interested.
00:14:50.240 --> 00:14:58.559
And all of that data is very valuable because if you're reaching out to the same person again in six months and they've already said that they're not interested, that is not a good look for the company.
00:14:58.720 --> 00:15:00.480
So all of that data is pretty valuable.
00:15:00.559 --> 00:15:03.919
That's your engagement data, CRM data, all of those kind of things.
00:15:04.159 --> 00:15:06.879
The other thing is your search preferences.
00:15:07.120 --> 00:15:12.639
So the way our agents work is that they don't just like do a one-off job and then forget about it.
00:15:12.799 --> 00:15:22.240
They actually retain important information about what kind of role, what kind of profiles your company actually likes and what kind of people perform better in your process.
00:15:22.399 --> 00:15:32.320
So the next time you're using juice box for another role, you have all of this context built in from like previous roles you've hired for that is incredibly valuable in running a more efficient process.
00:15:32.480 --> 00:15:35.519
Because otherwise you have to start from scratch and you have to calibrate things again.
00:15:35.679 --> 00:15:38.320
So I think both of those elements make the product pretty sticky.
00:15:38.799 --> 00:15:40.799
One actually follow-up on that as well.
00:15:40.960 --> 00:15:43.519
You mentioned efficiency of process.
00:15:43.840 --> 00:15:48.399
What are some of the other ways that customers define success using Juice Box?
00:15:48.559 --> 00:15:49.759
Is it truly just speed?
00:15:49.919 --> 00:15:51.600
Is it breadth of candidates?
00:15:51.679 --> 00:15:52.480
Now I know more.
00:15:55.360 --> 00:15:57.120
There's one piece, which is the speed.
00:15:57.200 --> 00:16:02.080
How much or
rather I think a better proxy is maybe time, like how much time are we saving you?
00:16:02.240 --> 00:16:10.080
How much less time do recruiters have to spend on the sourcing and engagement and more time that they can spend on closing candidates, on managing the process, those kind of things.
00:16:10.159 --> 00:16:11.440
So that's one dimension.
00:16:11.600 --> 00:16:17.120
The other dimension I like to think about is how important finding the right person for your company is.
00:16:17.279 --> 00:16:24.480
Because you're you're not just running a more efficient process, you're actually finding better people because you can spend time on it.
00:16:24.720 --> 00:16:38.879
So many of these recruiting processes, the way they go is they'll like start off, and then around like a one-month mark, when you're starting to get a little bit like concerned about like not having enough pipeline, the bar drops, and then you're not hiring as the kind of people you initially wanted to hire.
00:16:39.039 --> 00:16:41.039
That is what JuiceBox saves you from doing.
00:16:41.200 --> 00:16:52.399
So instead of like you having to actually run a very long and extensive process and finding people that are not even a fit, JuiceBox ensures that you have a good top of funnel of people that are actually a good fit.
00:16:52.559 --> 00:17:01.759
We make sure that you're engaging the right people, and that's one of the most high value things that a company can do is find the right people who can actually like steer things in the right direction.
00:17:02.000 --> 00:17:09.119
And it sounds like you've been moving rapidly deeper into the AI space, deeper into agents, and really automating the end-to-end recruiting process.
00:17:09.279 --> 00:17:13.680
But I'm wondering if we're sitting here in five years, what is the future of recruiting look like?
00:17:14.079 --> 00:17:28.720
Recruiting at the end of the day is if if you look at it in if you try to draw a box around it, it is a zero-sum game because there's a finite list of candidates, and those candidates are going to end up at some company and it's your job to win it, which is why it's incredibly competitive.
00:17:28.960 --> 00:17:36.880
The way we see recruiting evolve is that in let's say five years from now, choose box agents are going to be running the sourcing and the engagement piece.
00:17:37.039 --> 00:17:47.759
Choose box is gonna identify the right people for you, engage them, get them into your process, and talent teams will spend more and more time on providing better candidate experience and closing the right candidates.
00:17:47.920 --> 00:17:58.799
So spending more time with the candidates, making sure that you're running an efficient process that is more pleasant both for the candidates as well as the companies, and make sure that you're closing the right people and spending more time with them.
00:17:59.039 --> 00:18:02.079
How long does a typical search take when you're running it for?
00:18:02.240 --> 00:18:04.720
Maybe we can give an example role and just what it looks like.
00:18:05.039 --> 00:18:07.519
It's extremely variable from company to company.
00:18:07.680 --> 00:18:11.039
It depends really on like the kind of talent brand you have.
00:18:11.519 --> 00:18:25.359
But I would say like if if let's say you're looking for like senior software engineers in the Bay Area, that would and you're a series B company or a series C company, that would be a search that takes at least like one month to two months on the minimum.
00:18:25.440 --> 00:18:30.720
Like that's like a that's like an efficient process that you ran from like finding people to actually hiring them.
00:18:30.880 --> 00:18:32.400
And often it can take a lot more.
00:18:32.559 --> 00:18:40.400
It really just depends on how nuanced it's that role, how hard it is to hire for that role, and how much do people actually want to work at the company?
00:18:40.720 --> 00:18:42.720
Yeah, you actually bring up a very good point.
00:18:42.799 --> 00:18:50.880
We're all we're in San Francisco right now hosting this interview, and we all know and have heard about the talent wars that are happening in the AI space in particular.
00:18:51.039 --> 00:18:58.480
Given you're sitting at the intersection of AI and talent, curious, what are you seeing from a talent ecosystem in the Bay Area?
00:18:58.640 --> 00:18:59.680
How is Juice Box?
00:18:59.920 --> 00:19:05.440
I know you're powering over 3,000 customers, but you're also probably seeing how quickly talent is moving.
00:19:05.599 --> 00:19:07.279
Yeah, I would love to hear your perspective on it.
00:19:07.680 --> 00:19:22.079
It is a very interesting dichotomy, I feel like, right now, where on the one hand, if you look at some of the more macro picture reports, the amount of jobs is actually going down a little bit.
00:19:22.240 --> 00:19:23.680
So it's not on the up.
00:19:24.240 --> 00:19:31.279
At the same time, the top talent that people are looking for, that process is so much more competitive now.
00:19:31.440 --> 00:19:40.480
It's kind of unfortunate, but like a lot of the early career recruiting has gone down a little bit, where it's harder to find jobs, and it feels like that to a lot of early career candidates.
00:19:40.720 --> 00:19:50.240
Whereas at the same time, if you're a mid-level uh engineer or you're a senior level engineer in the Bay Area, people like are absolutely harping on those right now.
00:19:50.319 --> 00:19:52.319
Like they are so diverse effect to the market.
00:19:52.720 --> 00:19:53.039
Exactly.
00:19:53.119 --> 00:19:53.440
Yeah.
00:19:53.680 --> 00:20:03.680
And it's it's very interesting because I mean all all the tech stocks are still kind of fripping, and there's there's like so much capital, and it's it's really a good period for AI right now, which is why there's so much investments.
00:20:03.920 --> 00:20:16.079
That is why I think like the there's this like interesting dichotomy where like if you're
a mid-level or a senior engineer or anyone in the Bay Area who has some level of experience in the industry, then people are really going after those candidates.
00:20:16.400 --> 00:20:19.920
So we read an interesting stat, or maybe you told us, I forgot.
00:20:20.160 --> 00:20:24.160
But you reached your first million in ARR without hiring anyone.
00:20:24.400 --> 00:20:28.160
Do you think that is what the future looks like for AI companies being built?
00:20:28.559 --> 00:20:30.799
I think we took it to the extreme a little bit.
00:20:31.200 --> 00:20:31.920
And yeah.
00:20:33.200 --> 00:20:41.039
And for in our case, it was also an interesting piece because we we had this like long pivoting journey, and it's it's harder to move things around with more people.
00:20:41.200 --> 00:20:43.279
Like it's harder to steer a ship that's bigger.
00:20:43.440 --> 00:20:54.480
So I think it gave us a lot of flexibility in the early days to make sure we really double down and like really, you know, figure out exactly how we want to like shape things out and build our vision before like starting to scale the team.
00:20:54.720 --> 00:20:58.079
So in our case, yes, it was like a lot more extreme than it usually is.
00:20:58.319 --> 00:21:00.880
But I do think companies will be leaner and move faster.
00:21:01.039 --> 00:21:02.640
Like we see that within our engineering team.
00:21:02.799 --> 00:21:10.319
To date, our engineering team is now what, like six or seven people, which is like very little for the revenue stage and the number of customers we have.
00:21:10.480 --> 00:21:16.000
And the reason we are able to work efficiently is because all of these LLM tools, like they make you so much more efficient.
00:21:16.160 --> 00:21:25.680
You have to spend less time on writing code that LLMs are good at writing, and you can spend more time actually architecting systems and making sure you're you're you know making the right engineering decisions.
00:21:25.839 --> 00:21:32.559
So I think teams will continue to be more lean and we'll probably see more billion dollar companies being built with like less people.
00:21:32.960 --> 00:21:43.039
I think we're both observing that for sure from the investor lens of just like meeting companies that are at much later maturation from an AR perspective and much leaner teams.
00:21:43.519 --> 00:21:49.359
But you bring us to a pretty interesting point around fundraising and financing a business that's growing as quickly as yours.
00:21:49.920 --> 00:21:52.079
We know that you recently raised a series A.
00:21:52.160 --> 00:21:59.599
Maybe tell us a little bit about the raise and maybe what were some of the things you were focused on from a milestone perspective to say, hey, we're ready to take on.
00:22:00.160 --> 00:22:01.759
More financing and go even faster.
00:22:02.240 --> 00:22:07.759
We announced series A, which was 30 million, along with our which was 10 million, in the same announcement.
00:22:08.079 --> 00:22:11.920
We raised a lot later than most companies raise a series A.
00:22:12.079 --> 00:22:15.359
The reason we did that is because honestly, we didn't need the capital.
00:22:15.599 --> 00:22:17.279
We we're not burning any money.
00:22:17.440 --> 00:22:21.440
We're growing quickly, and we didn't feel like we we even needed to raise.
00:22:21.599 --> 00:22:27.759
But we ended also ended up realizing that right now is a very good opportunity because talent is so competitive.
00:22:27.920 --> 00:22:30.880
Everyone wants to hire good people to run an efficient process.
00:22:31.039 --> 00:22:33.759
LLMs can surely help with all of that process.
00:22:33.839 --> 00:22:41.440
So it was a very good time for us to like scale the company quickly, make sure we were able to grow even faster, which is why we ended up deciding to raise the capital.
00:22:41.920 --> 00:22:47.839
Even though we didn't need it at the time, we we felt like this can help us invest even further in growth and do things faster.
00:22:48.160 --> 00:22:54.400
I'm sure given that you did choose to raise later, there was a lot of interest in the round and uh sure a lot of preemptive interest as well.
00:22:54.559 --> 00:23:04.160
When you were thinking about who you wanted to work with, and presumably this might have been the first board seat, what type of investor or board member were you looking for?
00:23:04.319 --> 00:23:05.440
What traits really mattered?
00:23:05.519 --> 00:23:09.039
Aaron Powell There's two things we talk about internally a lot in terms of culture.
00:23:09.279 --> 00:23:14.400
One is moving fast because it's a competitive space, you need to do things quickly and it trade quickly.
00:23:14.559 --> 00:23:23.680
And the second we think we talk about a lot is intellectual honesty, which is this idea of like truth seeking, that we don't just want to believe things because they sound comfortable.
00:23:23.839 --> 00:23:29.440
We want to question it, we want to get to the bottom of it, and we only want to believe things if they're deeply rooted in truth.
00:23:29.599 --> 00:23:31.839
And that's what we are trying to look for in a partner as well.
00:23:31.920 --> 00:23:48.640
Is like, who can we find who actually shares our beliefs, who actually shares our values, who tries to question things from the ground up, question things from first principles, think about things more logically, follow a good line of reasoning, and honestly, like give it to us straight.
00:23:48.720 --> 00:23:55.519
Like someone who will be honest and upfront about things and not try to like, you know, bullshit you into believing like different things that are not not really true.
00:23:55.920 --> 00:24:04.240
Well, given you've brought on more capital with the Series A, it sounds like you guys have not slowed down at all with anything, you're just accelerating more.
00:24:04.480 --> 00:24:06.960
How are you thinking about further product expansion?
00:24:07.039 --> 00:24:09.599
What does
the roadmap look like for you guys?
00:24:10.480 --> 00:24:15.039
We're working very actively on a memory layer right now, which is what I talked about earlier as well.
00:24:15.200 --> 00:24:18.000
Is like how can we retain more and more context from our agents?
00:24:18.160 --> 00:24:26.640
We see companies, like a lot of our contracts, the way they look like is they will buy 20 seats and then they will buy 20 agents, one for every single one of their recruiters.
00:24:26.799 --> 00:24:33.599
So the recruiters will use the platform, but alongside they will spin up all of these agents that will do a lot of the sourcing for them.
00:24:33.839 --> 00:24:40.480
We're generating a lot of data every single day because you're spinning up all of these agents that are like going out there finding candidates for you.
00:24:40.559 --> 00:24:50.480
And they have all of this information about your preferences, about the kind of people you like, about the kind of people that are making it through your process, who actually end up getting hired, and who's even interested in the first place.
00:24:50.720 --> 00:24:57.519
And all of this data is incredibly valuable to shape your talent organization over the next few years, even.
00:24:57.920 --> 00:25:03.440
Because all of this data helps you build on top of whatever experiences your team has already had.
00:25:03.599 --> 00:25:18.960
So, one thing we are working on a lot is building a more long-term memory layer and making it even more transparent to the user about what the AI is learning from you, how is it maintaining a memory set in the background, and how it is starting to actually use that in more future roles.
00:25:19.200 --> 00:25:23.519
Another thing we are very excited about is more deeper integrations with ATSs.
00:25:23.759 --> 00:25:31.200
So some of these enterprises use platforms like Workday and Oracle, and they have data from the last few decades, like not even years.
00:25:31.519 --> 00:25:35.359
It's literally like 30 to 40 years of applicant data.
00:25:35.599 --> 00:25:45.200
And all of this data is also incredibly valuable in shaping your recruiting strategy because you know who's applied, you know who went through the process, you know who's a successful employee at your company today.
00:25:45.440 --> 00:25:51.519
So we're trying to capture more and more insights from that data and make sure we are able to more deeply integrate with these systems.
00:25:51.680 --> 00:25:58.240
And our vision is to go from a job description to a first call booked with a qualified candidate without any human intervention.
00:25:58.400 --> 00:25:59.759
And I think we're working towards that.
00:26:00.079 --> 00:26:02.720
When do you think that will be actionable?
00:26:03.039 --> 00:26:05.200
I think it's a lot closer than most people think.
00:26:05.680 --> 00:26:07.119
In a few months, we'll be seeing that.
00:26:07.200 --> 00:26:10.400
We already see that with some roles that a lot of our customers hire
for.
00:26:10.480 --> 00:26:13.119
And I think we'll start seeing that across more and more roles very soon.
00:26:13.359 --> 00:26:13.599
Awesome.
00:26:13.680 --> 00:26:16.000
I guess my last question on the memory layer, what does it look like?
00:26:16.160 --> 00:26:19.599
Are you using, are you leveraging the off-the-shelf open source software?
00:26:19.839 --> 00:26:22.799
Are you leveraging OpenAI or how are you embedding it?
00:26:23.119 --> 00:26:28.079
We use a lot of like foundational models directly from the labs and use a lot of APIs.
00:26:28.319 --> 00:26:31.279
We do have some fine-tuned models that we self-host as well.
00:26:31.359 --> 00:26:34.079
So we use some of the open source models from DeepSeek.
00:26:34.480 --> 00:26:38.720
But the most amount of success we've actually had is with the foundational models from the labs.
00:26:38.880 --> 00:26:41.119
So we mostly rely on that kind of infrastructure.
00:26:41.519 --> 00:26:43.440
Well, do you want to hop into a quick fire round?
00:26:43.759 --> 00:26:44.640
Yeah, let's do it.
00:26:44.960 --> 00:26:45.599
Awesome.
00:26:45.839 --> 00:26:48.880
What is one word to describe the future of recruiting?
00:26:49.279 --> 00:26:49.920
Automated.
00:26:50.960 --> 00:26:53.680
What is a piece of advice for first-time founders?
00:26:54.079 --> 00:26:55.039
Stay persistent.
00:26:55.119 --> 00:26:56.559
You can pivot your ideas.
00:26:56.640 --> 00:26:58.319
That's like totally fair.
00:26:58.640 --> 00:27:03.599
And you just have to keep going, get ready for like a world of pain and be okay with it.
00:27:04.000 --> 00:27:08.079
What's the hardest decision you've had to make so far as a first-time founder?
00:27:08.480 --> 00:27:16.559
Changing directions was extremely like because the the what you're really saying when you're pivoting is that everything I did over the last few months is going out of the window.
00:27:16.640 --> 00:27:19.279
It's completely useless, and I'm starting from scratch.
00:27:19.519 --> 00:27:26.480
And saying that is extremely scary, but it's also very, very necessary because you don't want to go down if that does not scale.
00:27:26.640 --> 00:27:30.240
So And what can you not automate today that you can in the future?
00:27:30.640 --> 00:27:31.279
Pushback.
00:27:31.440 --> 00:27:37.440
Whenever you actually reach out to a candidate, sometimes they're interested, sometimes they're not interested, and sometimes they are on the fence.
00:27:37.599 --> 00:27:43.519
And when they're on the fence, they need to be sold to, and they you need to answer questions about whatever their concerns are.
00:27:43.759 --> 00:27:50.079
Handling that objection and getting them onto a first call is something I strongly feel like we can automate soon.
00:27:50.240 --> 00:27:54.240
And then there's a whole scheduling piece after that of like getting them on the calendar, booking a call.
00:27:54.400 --> 00:27:56.079
So all of that I think we can also do.
00:27:56.480 --> 00:27:57.920
I have one more bonus question.
00:27:58.160 --> 00:28:01.839
Given that I think most of our listeners know that it's a very active space right now.
00:28:01.920 --> 00:28:04.160
There's a ton of competitors popping up.
00:28:04.400 --> 00:28:12.640
And if I'm coming to you as an enterprise company, why should I choose to work with JuiceBox versus the next AIHR tool?
00:28:13.039 --> 00:28:21.279
The first thing is social proof that you should look at our list of customers and you should see that we have 3,000 customers and we have some of the best companies in the world using JuiceBox.
00:28:21.519 --> 00:28:31.119
And the reason they might made that decision is because they did exactly what you're planning to do, which is try three or four different tools and end up realizing that they're getting the most gain from JuiceBox and end up going with us.
00:28:31.279 --> 00:28:32.960
So the social proof is the first thing.
00:28:33.119 --> 00:28:38.960
And the second piece is that we approach things from a very, very first principles perspective.
00:28:39.039 --> 00:28:45.279
Like we're not trying to replace your ATS, we're not trying to like shift around your existing workflows.
00:28:45.359 --> 00:29:03.359
We're trying to deeply integrate with it and make sure that we can save you the most amount of time and make sure that your organization is competitive and not just because we want to sell you another piece of software that's like N plus one, like just slightly more efficient, but because we actually want to save the grant work that people don't enjoy right now.
00:29:03.759 --> 00:29:04.240
Awesome.
00:29:04.640 --> 00:29:05.359
That's amazing.
00:29:05.680 --> 00:29:06.240
Thanks so much.
00:29:06.319 --> 00:29:06.720
This was great.
00:29:07.039 --> 00:29:07.279
Of course.
00:29:07.440 --> 00:29:08.319
Thank you so much for having me.