November 4, 2025 • 48 mins
Michel Tricot is the co-founder and CEO of Airbyte, the open-source data movement platform he launched in 2020. Before Airbyte, Michel led integrations and served as Director of Engineering at LiveRamp, where he scaled the teams and pipelines that synced massive data volumes. He also helped build rideOS as a founding engineer and Director of Engineering. Michel has spent 15+ years in data infrastructure, with a focus on commoditizing data pipelines and giving teams control and sovereignty over their data.
Discussed in this episode
- Why Airbyte launched open source first (catching engineers “at the search”)
- Project-market fit vs. product-market fit, and why they’re different
- The content engine: founder-led writing, shipping slides, and radical transparency
- Turning interest into community: 25k+ Slack, champions, and hiring from within
- The near-misses: hiring ahead of PMF, support-heavy community, cloud complexity
- Going upmarket: enterprise motion, longer cycles, and team ramp realities
- AI wave → agents as “data consumers” and what it means for pipelines
- Replatforming for control & sovereignty, not just “more connectors”
This episode is brought to you by our sponsor: ZoomInfo
ZoomInfo is the GTM Intelligence Platform built for sales, marketing, and RevOps.
By unifying data, workflows, and insights into a single system, ZoomInfo helps revenue teams find and engage the right buyers, launch go-to-market plays faster, and drive predictable growth. With industry-leading accuracy and depth of data, it gives your team the intelligence advantage to win in competitive markets.
It’s trusted by the fastest-growing companies and has become the category leader in GTM Intelligence.
Learn more at zoominfo.com.
Follow Michel Tricot
- LinkedIn:https://www.linkedin.com/in/micheltricot
- X (Twitter):https://x.com/MichelTricot
- Website:https://airbyte.com/Airbyte
Host links
- LinkedIn: https://www.linkedin.com/in/sophiebuonassisi/
- X (Twitter): https://x.com/sophiebuona
- Newsletter: https://thegtmnewsletter.substack.com/
- Website: https://gtmnow.com/
Where to Find GTMnow
- Website: https://gtmnow.com
- LinkedIn: https://www.linkedin.com/company/gtmnow
- X (Twitter): https://x.com/GTMnow_
- YouTube: https://www.youtube.com/@GTM_now
- Podcast Directory: https://gtmnow.com/tag/podcast
The GTMnow Podcast
The GTMnow Podcast is a weekly podcast featuring interviews with the top 1% GTM executives, VCs, and founders. Conversations reveal the unshared details behind how they have grown companies, and the go-to-market strategies responsible for shaping that growth.
Visit gtmnow.com for more episodes and other interesting content.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
SPEAKER_00 (00:00):
People are willing to put time into the project and
the product that we arebuilding.
How do you actuallycommercialize it?
It's a different story.
And to me, that's what PMFactually is, where everything
goes super fast, every deal getsclosed in like a week, two
weeks, one month max.
SPEAKER_03 (00:15):
You launched in 2020.
Now you're valued at over abillion dollars.
Michelle didn't play the typicalfast launch playbook.
He went open source first,community first, and GitHub
first.
And it ignited one of thefastest bottoms-up adoption
curves in modern datainfrastructure.
Today, AirPyte is valued at overa billion dollars, powering data
movements for thousands ofteams, including over 20% of the
(00:37):
Fortune 500.
And they got there in a reallyinteresting way.
They built in public, compoundedthrough community, and turned
contribution into a distributionmode.
In this conversation, we breakdown the stories and lessons
behind all of this growth,including a really important
lesson on separating productmarket fit from project market
fit.
All right, let's get into it.
(01:09):
Michelle, welcome to thepodcast.
SPEAKER_00 (01:11):
Thank you for having me.
Great to be there.
SPEAKER_03 (01:13):
It is a pleasure.
And it hasn't been long since wesaw each other earlier this
week, in fact.
So it's great to see you again.
SPEAKER_00 (01:19):
Yeah, no, that was a good, a good event.
Like that was the Tech Crunchone that was very solid.
SPEAKER_03 (01:25):
Yeah, that was great.
And your your uh session washighly attended.
Sounded fantastic.
Exciting to pick your brain alittle bit more intimately than
at the event itself.
And you launched in 2020.
Now you're valued at over abillion dollars.
Take us back.
We want to know the how behindthis type of growth.
SPEAKER_01 (01:43):
Yeah.
SPEAKER_03 (01:44):
And before we even get started from the beginning,
a lot of our audience aren'tengineers.
A lot of operators, a lot offounders.
Give us a high level of Airbite.
What does Airbite do?
Kind of the two-liner foreveryone listening.
SPEAKER_00 (01:57):
Yeah.
So AirBite is an open datamovement platform, meaning that
we can take any pieces of dataacross any system and we can
deliver it into a place where itwill deliver value.
So a very strong use case isgoing to be everything related
to analytics.
How do you go across yourcompany, look at all the
services that you have, all thedata sources that you have, all
(02:19):
the silos that you have, and howdo you make it seamless to move
that data into warehouse so thatyour analytics team can actually
extract insight from it and makedecisions from it.
And that's really how westarted.
There is a ton of use case whenit comes to moving data.
You know, we're talking aboutagents these days, is like how
do you get the data into agents?
So that's very much what thevery high-level value of Airbite
(02:44):
is.
SPEAKER_03 (02:44):
Super helpful.
And now let's go back to thebeginning.
You launched on GitHub.
Why open source as opposed to atraditional product launch?
SPEAKER_00 (02:53):
Yeah.
So when you're thinking about,let's take the analytics use
case as an example.
You go from like the outcome youwant to drive, which is I want
to be able to understand mybusiness.
The first thing you think aboutis okay, I will need to have
dashboards, I will need to havea team, I will need to have a
warehouse.
(03:14):
And the moment you have thesetwo, what you realize is that
you also need the data,obviously.
SPEAKER_02 (03:18):
Yeah.
SPEAKER_00 (03:19):
And this is a very organic um behavior from people,
which is it's not thoughtthrough so much as a strategy,
but more as an enabler.
So they're gonna go bit by bitthinking, oh, I need this
particular silo, I need thisparticular silo.
And it is very hard to actuallythink about the pain that it
(03:43):
will be if you build ityourself, or it will be very
hard also to find platforms thatcan support every single silos
that you have.
And for us, when we did opensource, what we wanted is to go
and talk to the team that arebuilding all these different
connectors.
So when you're an engineer andyou're being asked, oh, I need
(04:06):
Stripe data to be in thewarehouse, the first reflex that
an engineer will have is goonline, check how do I move data
from Stripe, Salesforce, HubSootor you name it, into my
warehouse.
And we wanted to catch thesepeople exactly at that time.
We wanted to provide them valuethe moment they have that little
(04:27):
painful script that they have towrite and give them something.
So open source at that point isgenerally the best solution
because I mean I'm an engineer,I'm a little bit lazy when it
comes to if I can avoid buildingsomething, I will.
SPEAKER_02 (04:42):
Yeah, fair.
SPEAKER_00 (04:42):
And open source is generally the solution for that,
and that's really why we wentfor like open source.
The other reason is there's aninfinity of places where data
can be.
So it is impossible for a singlecompany to make a product that
will address all the long tailsof data connectors.
What we need is, and what thecommunity needs is like, in a
(05:05):
way, all working together in agoal of like addressing all
these use cases.
And that's why open source forus was a solution.
Like you, you know, you canthink about the Linux kernel.
Well, all the drivers are beingbuilt either by the community,
either by by vendors, but theLinux project is not building
all these drivers.
(05:26):
They are asking the community tobuild those, and that's how you
just get to the best uh uhproduct on the market.
SPEAKER_03 (05:34):
And it feels like we're seeing more and more
companies open source.
Do you feel that also?
SPEAKER_00 (05:39):
Yes.
Um, yes, and I think it'sbecause the technology,
especially this, you know, opensource is very, very present in
AI, for example, because thereis almost like a complete stop
of the old world versus the newworld.
Like everything has to bereinvented.
And people who are makingdecisions today have to catch up
(06:03):
on a lot of context.
So, what they do is actuallythey go talk to their team and
ask them we I we need to createan agent for this particular use
case.
What technology should we beusing?
And open source generally worksreally well with technical
profiles.
And I think that's one of thereasons.
There are also a lot of thingsaround sovereignty and control
(06:25):
that comes with open source andalso future proofing because you
can always update the projectyourself if you want to.
And to me, that's a directionthat we're seeing.
And having a community thatbacks a project just you cannot
beat that velocity.
SPEAKER_03 (06:42):
Yeah, so true.
So true.
And okay, so you launched in2020.
When you uploaded the repo, didyou know that it would take off
the way that it did?
SPEAKER_00 (06:53):
No, we didn't know.
We are so in the story ofHairbyte, like Airbyte started
really just two months beforeCOVID really hit the world.
SPEAKER_03 (07:02):
What a time to start.
Yeah.
SPEAKER_00 (07:04):
And we had an initial product at the time,
which was also related to dataintegration, but more geared
toward marketing teams.
And what happened with COVID isboom, all the marketing team got
frozen, laid off, etc.
etc.
Because company had to figureout, okay, what does the world
look like now?
And you know, as a founder, youput your life into uh a company,
(07:27):
into building a product, and youdon't want to be a vitamin that
I like to joke about that is notgoing to survive a global
pandemic.
So what we did is we actuallywent back to the drawing board.
And in July, like during theperiod of like March to July, we
were building prototypes, etc.
etc.
But but we're also talking a lotwith the audience that we wanted
(07:49):
to build a product for, whichwas data people.
And all these people, they werealways having a solution that
they would buy, a solution thatthey will build, another
solution that they would build,another solution that they would
buy.
So it was like a collection oftools everywhere just to move
data.
And what we've done is justkeeping in touch with all these
(08:10):
people and keeping them in theloop of what we were building,
what product.
SPEAKER_03 (08:15):
What are you doing that?
How are you keeping them in theloop?
SPEAKER_00 (08:18):
So at the time during COVID, everybody, I think
a lot of people were veryavailable on LinkedIn.
SPEAKER_01 (08:23):
Yeah.
SPEAKER_00 (08:23):
So we're very, very active on LinkedIn.
So we were always trying to talkto the right people, going on a
Zoom with them for like 15minutes, 30 minutes, and then we
would ask them, Do you want tobe following what we're doing?
And say yes.
And then we created the firstmailing list that we had, and
every time we had updates, wewould just say, Oh, this is what
(08:44):
we're building.
If you want to, we can give youa quick demo of what it looks
like, and you can give usfeedback.
That was before we published therepo.
And I think it was in Novemberwe actually put the um the repo
out.
And suddenly, first of all, likethis initial group of people
started to download the thesoftware, started to give us
(09:06):
like real feedback, and fromthere it just went uh in hockey
stick.
SPEAKER_03 (09:11):
Yeah, incredible.
How did you feel just seeingthat growth after you said it
yourself when you're a founder?
You put you put everything intoa company.
SPEAKER_00 (09:19):
Yeah.
It's uh I felt very good in away, which is people are willing
to put time into the project andthe product that we're building,
and yet it is super immature.
And you know, we always talkabout PMF in the the founder
founding sphere.
PMF, my definition, having seenthat, is it's when people are
(09:40):
willing to go above and beyondto make something that is not
yet mature, that is not yetworking, and they are willing to
put the effort to make it workbecause it is solving such an
intense problem for them thatthis little pain of making it
work is better than the big painof having to do it yourself.
(10:01):
Um, and yeah, it felt good.
After that, yes, I knew that thetechnology needed to become
better, but you have to launch.
SPEAKER_03 (10:08):
Yeah, yeah, exactly.
Usually, if you're at a pointwhere you feel like it's good
enough, it's too late from alaunch perspective.
SPEAKER_00 (10:15):
Exactly.
Like you want to get thefeedback as fast as possible.
You just want to build what isactually going to deliver value
for your community.
SPEAKER_03 (10:22):
A quick pause to tell you about a company you
need to know.
ZoomInfo is the go-to-marketintelligence platform built for
sales, marketing, and rugoffs.
By unifying data, workflows, andinsights into a single system,
ZoomInfo helps revenue teamsfind and engage the right
buyers, launch go-to-marketplacefaster, and drive predictable
growth.
With industry leading accuracyand depth of data, it gives your
(10:42):
team the intelligence advantageto win in competitive markets.
It's trusted by the fastestgrowing companies and has become
the category leader ingo-to-market intelligence.
Learn more at zoominfo.com.
So, how long did it take to hitPMF in your definition project
market fit?
SPEAKER_00 (11:01):
I'm actually splitting it because there are
two paths in the life ofAirbytes.
There is what I call projectmarket fit, which is we managed
to create a project that wasvery much resonating with an
audience, data engineers, dataanalysts, etc.
And they were just taking theproject and using it and
(11:22):
contributing to it.
Product market fit for me alsocomes when you start pulling the
foundation also of uhmonetization.
And this is a different storybecause it's easy to take a
product from GitHub.
How you actually commercializeit, it's a different story.
And to me, that's what PMFactually is.
So I would say open source wasproject market fit.
SPEAKER_03 (11:45):
Got it.
Okay.
Well, take us through a littlebit of the evolution then.
Because you got project marketfit.
What kind of go-to-marketdecisions did you make along the
way that helped you to get toproduct market fit from project
market fit?
SPEAKER_00 (12:00):
We were not using regular channels.
It was all about content.
I mean, at the time, contentmarketing was a thing, but I
don't think it was as uh aspopular as it has been like in
2023, 2024.
But we were just always pushingarticles, giving details about
how the what the company isdoing, what the project looks
(12:22):
like, and just getting people tobe part of our adventure.
And that created trust, thatcreated curiosity, that created
a lot of awareness.
You know, we published ourfundraising slides, for example.
So that was a way for us of likeengaging the community into what
we are doing.
So that uh that to me issomething that not a lot of
(12:43):
people have done in the past.
So it was very uh I think veryI'm I'm pretty proud that we've
done that.
It's very, very innovative.
And then, yeah, like we'vealways been very strong on
content, engaging with thecommunity.
SPEAKER_03 (12:57):
So but it is time consuming content.
So how do you think about thatas a founder?
How do you balance yourschedule?
When do you work it in?
What's your actual cadence orwas at the time for any kind of
founders or operators listeningthat are looking to up level
their content?
SPEAKER_00 (13:11):
Yeah.
It is time consuming, but youknow, if it's working and you
feel it's gonna be workingbetter than any other solution,
you just continue and you youexploit that uh that channel as
much as you can.
Um after that, we yeah, we arenow we're doing we continue to
(13:31):
do a lot of content, but we arealso a lot more like traditional
channels like ads, uh SEO, GEO,etc.
etc.
SPEAKER_03 (13:39):
So yeah.
And did you write it allyourself?
Did you hire a ghostwriter?
Like when did you actuallyphysically put kind of uh pen to
paper, if you will, or fingersto keyboard?
SPEAKER_00 (13:51):
I would say the first year and a half, it was my
co-founder and I writing.
The team also was writing.
So we really created that cutethat internal culture of let's
write something.
My VP of engineering wrote anamazing article about the pain
of building connectors that wekeep referring to, even five
years later, uh, because itreally explains the pain.
SPEAKER_03 (14:14):
Yeah.
Well, it's funny too.
Five years later, and the painis still the pain.
SPEAKER_00 (14:18):
The pain is still the pain, and uh yeah.
SPEAKER_03 (14:20):
Yeah.
Incredible.
Okay.
So you lean into content early,and that helped sounds like
create a bit of a tribe, and youhave a very strong following of
people that are passionate aboutthe product and the space and
the solution that you built.
How did you think about actuallytaking that interest generated
from your content and othermeans and turning it into more
of a community motion?
SPEAKER_00 (14:41):
Yeah.
For me, at that point, so we wealso created a Slack community
at the time.
I think today we have about25,000 people on it.
And the way we created thecommunity was in twofold.
One is we were helping thecommunity a lot.
(15:02):
We are doing a lot of supportbecause we are building the
platform.
And so every time someone had anissue, that was a product
feedback for us.
So we spent a lot of time in2020, like end of 2020,
beginning of 2020, like all of2021, and uh we we continued
after, but that was very, veryintense, a year and a half,
where we were always on Slack.
(15:24):
Every single issue that wasreported, we would just have
something shipped the day afteror like the week after.
So I think that created that'sthat was one thing that created
that helped uh building thecommunity.
And then what we did is we alsoidentified a lot of champions
within the community, likepeople that wanted to help other
people.
(15:45):
And yeah, we really engaged withthem.
We actually hired one of thefirst community managers that we
that we've had at uh at Airbite,is someone that we actually
brought from the from thecommunity that was he started to
build an airflow connector, likean airflow integration, and say,
oh man, that's amazing.
And we didn't ask him anything,and at some point we asked him,
(16:05):
like, do you want to uh to do ityour full-time job, like to
engage with the community, writecontent, etc.
etc.
And they're like, Yeah, let's doit.
SPEAKER_02 (16:12):
So it's amazing.
SPEAKER_00 (16:13):
That to me is like the community engagement is is
absolutely key.
That's how you create thattribe, that's how you create
that snowball effect.
It's it's not something that youput on, you say, I'm building
community, and it's gonna happenby itself.
No, it is something that has tobe worked on, and you have to be
intentional about what you wantto do uh with the community.
SPEAKER_03 (16:34):
If you were to look back now with the benefit of
hindsight, it sounds likecommunity and content are two
pillars that helped with yourgo-to-market motion.
Yep.
Is that correct?
And also are there other pillarsthat you'd say were really
pivotal in your growth?
SPEAKER_00 (16:49):
Um we did a lot of events, actually, very
specialized events around uhdata, whether they were open
source events or like Snowflakeor Databricks events, it's just
always getting where the peoplewere.
And that was something thatworked pretty well.
(17:11):
It allowed us to get a lot ofpeople, new people interested,
or just to engage in real lifewith people.
Yeah um yeah.
I would say, and here's reallywhat happened between 2020 and
2023.
After that, we had we added afew other things on top of uh
like how we engage with thecommunity, etc.
(17:33):
etc.
But that to me was very verymuch like the three pillar of
what we've done.
It's like giving a window intothe company to people, giving in
a window into how the theengineering team is building,
giving a window into everythingwe're doing.
SPEAKER_03 (17:49):
And do you still operate that way?
SPEAKER_00 (17:51):
Uh a little bit less.
Um but we continue to have thatconstant engagement with uh with
people.
Like, you know, when the thegreat thing that when you have a
community like that is someonein whether it's a customer,
whether it's uh it's a user, isgoing to ask you a question or a
(18:15):
feature, and then you're it'sgonna go into your head and say,
okay, is that really useful?
Or is it just for that person?
And so what you do is you go inyour on your community and you
you just post a very simplequestion, like is that something
that resonates with you?
And in 30 minutes, you have likehundred people that are
(18:37):
replying, yes, no, yes, but inthat way.
So it really accelerates how youdo product discovery, how you do
uh product development.
So that's uh that's more likehow we've changed a few things
along the way.
It's like we're we're leveragingthe the community a lot more for
like what new features we shouldbe building rather than really
the the the core valueproposition of the ambite.
SPEAKER_03 (19:00):
Right.
It's uh a feedback loop, yeah,essentially.
SPEAKER_00 (19:03):
Yeah.
SPEAKER_03 (19:03):
Great, and a very, very rapid one too.
SPEAKER_00 (19:06):
Very rapid one.
SPEAKER_03 (19:07):
So content, community events, pillars that
you did incredibly well to reachthe point you are now.
There's always the other side ofthe story of you know, what were
the the areas that didn't quitehit as well or almost the
near-death experiences along theway that every startup goes
through?
SPEAKER_00 (19:22):
Yeah, so as I said, like the beginning of the of how
we are engaging with thecommunity was very a lot of
support, like helping them besuccessful with the product.
And there was this moment whereeven in our in how we were
working, our community becamevery much of a like support
channel rather than likebuilding a uh a community that
(19:49):
was just helping each other.
Um, and that to me was uh issomething that we could have
been more intentional at thebeginning around how do we um
how do we get to like communitymembers helping each other,
community members like meetingeach other outside of just like
rather than becoming a very muchlike support-oriented um uh
(20:13):
community.
And the thing is, once thishabit is taken, it's very hard
to shift uh into a differentdirection.
I think we succeeded, but ittook us a lot of time.
We should have been moreproactive thinking about okay,
the community is amazing, butwhat is the future?
Like, how do we make it morevibrant, more um yeah.
(20:33):
How do we create a community ofprofessional that work in data
and that are just gonna learnfrom each other and not just
from us?
SPEAKER_03 (20:41):
Mm-hmm.
Yeah, it completely makes sense.
It's kind of the the the tellall tales, the tell-old tale
story of community is a lotharder in practice, and it does
require some really deepintentionality around fostering
it.
SPEAKER_00 (20:55):
It does, it does.
SPEAKER_03 (20:56):
Yeah, and what does that team kind of composition
look like right now at Airbite?
SPEAKER_00 (21:01):
Um so we have we have a we have a DevRel person,
and this this person is more umfocused on the like the content
strategy geared, oops, gearedtoward the community.
And we have a community manager,meaning someone that just
engages, identifies champion,uh, gives them access to um
(21:26):
early features, etc.
etc.
And we also have people um ininternally we call them like
customer engineering, wheretheir focus is to make sure that
every product feedback aroundconnectors is being funneled
through the team to make surethat our connectors keep getting
better and better and better.
(21:47):
So this is more like for thecontributors of the platform.
So we really have a differencebetween like the users of the
platform and the contributors ofthe platform, and we handle
these two groups differently.
SPEAKER_03 (21:58):
Gotcha, gotcha.
Okay.
What are some other areas thatyou know along the journey,
again, reflecting back, havejust been some of the most
pivotal things that maybe youdon't you don't see or talk
about as much?
SPEAKER_00 (22:11):
Um I think that was the the realization of why are
so many companies using airbite.
Is it just connectors or is itsomething else?
And connectors is a is levelone, but there is a a second
level to it, and it took us alittle by a little bit of time
(22:33):
to figure it out, is people werealso using data, uh airbite,
because there was so much redtape around the data that they
had internally, that having aplatform that they fully control
that runs within theirinfrastructure, it's a byproduct
(22:55):
of open source.
And we did not realize uh Iwould say like fast enough that
that was one of the key reasonswhy so many teams were adopting
airbytes.
So, you know, when we started toto do the airbyte monetization,
we said, okay, we're gonnafollow the we're gonna skip the
step of doing support for peoplethat are deploying airbytes, and
instead we're gonna go directlyto a cloud product.
(23:17):
And very quickly we realized,yes, cloud is getting traction,
but we are not able to convertevery person that is using
airbite to using airbite cloud.
And at that point, we just wentback to the drawing board,
started to talk to them, andthat's when we discovered that
in that case, like productmarket fits was not just
connector.
It was the fact that these pipeswere under their control.
(23:40):
And that was a big, a big thing,and I would say we we wasted a
little bit of time on trying tobuild something fully cloud when
what people needed was controland sovereignty.
SPEAKER_03 (23:53):
Got it.
Okay.
SPEAKER_00 (23:55):
More like, you know, when you're searching for PMF,
it's not a straight.
SPEAKER_03 (24:00):
Never linear, never linear, no.
And what are you most excitedabout thinking now forward?
SPEAKER_00 (24:06):
Yeah.
Well, you know, every every timeI hear about how do I make I
mean to me like the AI wave thatis happening right now is just
one of the most exciting thingsfor me and for the for the
company.
Like analytics is very much acore part of what we're doing,
but we're getting so much pullinto different types of data
(24:30):
access.
And that is something that we'retoday encoding into the platform
and into our connectors.
It's not just humans consumingdata today.
Yeah, it's agents that candiscover what's available,
discover what it looks like, andmake decisions.
So, yes, the technology is notyet completely mature on either
side, whether it's airbite,whether it's like agency
(24:50):
platform, etc.
etc.
But you can see how fast it'smoving, and I think it's very
energizing, especially in theinfrastructure world, to see
that that energy being uh beinginjected.
So that yeah, I I I talk aboutit all the time.
SPEAKER_03 (25:06):
So Yeah, no, that's fantastic.
And I mean you mentioned that atthe very beginning around how
now it's agents consuming thistype of data.
How does that transition in theoverall industry?
What does anyone need to knowabout what this transition
actually means?
SPEAKER_00 (25:22):
Yeah.
You need to you need to forgetabout a lot of your existing
patterns.
You know, I was chatting with uhwith a CTO uh last week, and he
told me very bluntly, I don'tknow, maybe he was trying to uh
(25:42):
to be a little bit uh uhprovocative here, but he said he
told me, Michelle, all thetechnical knowledge I had
stopped two years ago.
I had to fully reinvent myselfand reinvent my team.
Uh so yes, some things are stilltransferable, but your default
should always be thinking abouthow do I build in that new
(26:05):
world?
Is there a solution?
No.
Okay, maybe I go back and usethe techniques of the of the the
older world.
But that's really what what I'mseeing is people have to rethink
how they are doing their job.
Because one thing that ishappening in Teams is a lot of
people are using AI today toremove from their play the thing
(26:28):
that they don't like doing.
That's very easy.
Like people have a very stronguh willingness to stop doing the
things they hate doing.
So for that, like AI is is isamazing.
Like, you know, if you're anengineer like writing unit
tests, writing integrationtests, that's great, but that's
just level one.
The moment you actually startchanging your mindset is when
(26:49):
you're looking at the things youlike doing and how can you
leverage AI for those.
But those are hard because thethings you like doing are the
things also that can bring you alot of energy in your in your
day-to-day.
And those are the things thatpeople should really be focusing
on.
On okay, this thing that I'mdoing every day, I love doing
it, but can I do it using AI,using an agent?
(27:12):
Can I ask my engineering team tobuild an agent to solve that
particular problem?
Is there an AI product thatexists that can do it and
removes that from my plate?
And then I can focus on morethings and I can become faster.
But to me, it's really aboutreinventing um reinventing it.
For data, the way you accessdata is very different.
SPEAKER_01 (27:33):
Yeah.
SPEAKER_00 (27:34):
Um but just having a warehouse doesn't cut it.
Like you need to have like anagent does live processing, it
needs to have like little piecesof data here and there.
You need to provide access tothe agent in a different way.
So that's and that's what that'swhat we've been building.
SPEAKER_03 (27:54):
Yeah, absolutely.
How did that change your productroadmap overall?
Did you have like this crazymoment in a way where it was
like the realization that youentirely have to pivot?
Or is it gradual?
SPEAKER_00 (28:07):
I I wouldn't say it's a it's a pivot because it's
more like a an extension, butalso sometimes we like to talk
about replatforming, which iswe've we've built the plat the
platform for like a specific usecase in a specific course, but
there are new ones that arecoming that are going to pick up
massively over the next fewyears.
(28:28):
And we need to be thinking abouttaking all the learnings that
we've had here and how do wethink about replatforming it to
just have a larger breadth ofuse case.
So that's that's more how howwe're thinking about it.
Uh I don't know, I would say2024 is when we even even before
like summer 2023 is is when westarted to like tippito into it.
(28:50):
But 2025 is a moment where we wewent all in on that.
So we still have the theanalytics product, it's a it's
an amazing product, but we arereally building on top of that,
like leveraging part of it, butalso rebuilding a platform that
allows agents to uh to interactwith data.
So it's pretty pretty cool.
SPEAKER_03 (29:12):
Super cool.
And you're hitting the groundrunning, tons of growth, you're
hiring lots of folks on theteam.
Like, how do you think aboutdeveloping that team to take it
to the next stage of growth?
SPEAKER_00 (29:22):
Oh, you have to be hammering using AI every single
day, every single or-ens that Ido every Wednesday morning.
It's about putting the spotlighton new uh new way of leveraging
AI.
And not just the layer one,which is do the thing you don't
like doing.
It's really about how are peoplebuilding things that change
(29:46):
their, actually change the thedefinition of their job.
So well, if it's if it's onsales, it's gonna be around like
how do they do uh like discoveryof account, it's gonna be how
they connect um uh Likedifferent news together, how do
they connect to like pastconversation that's happened on
support?
(30:07):
So it's really about likeaggregating all this information
in one single place and havelike all the context available
to them at the right time.
On engineering, well, we talkedenough about engineering and how
agents are transforming thelives of engineers, and that's
what we've been doing atAirbytes for the yeah, for the
for the past year.
SPEAKER_03 (30:26):
Yeah.
And so it sounds like youdisseminate this information
internally.
You said weekly.
SPEAKER_00 (30:31):
Weekly.
SPEAKER_03 (30:32):
What does that look like?
Everyone's on a team callweekly, or how are you
spotlighting people?
SPEAKER_00 (30:36):
Yeah, so the whole company we'd spend like 30
minutes together and we go overlike some updates, but then we
always have like onepresentation that is just about
AI.
And myself, like I generallystart the whole hand and I
always have a few slides aroundlike the wins of the week.
SPEAKER_01 (30:52):
Yeah.
SPEAKER_00 (30:52):
And well, we have a channel on Slack where people
just write their wins and I justgonna pick one or two.
And that's why I mean likepulling the spotlights on
specific individuals that havedone something innovative with
it.
And I think that creates like agood dynamic.
Like people want to be on thewin slide, etc.
etc.
So it creates a little bit oflike internal competition.
SPEAKER_03 (31:14):
Yeah, internal competition and also ideation.
Exactly.
I find sometimes the biggestblocker is the inspiration and
ideation around like what can Iactually do with AI?
SPEAKER_00 (31:23):
Exactly.
SPEAKER_03 (31:23):
So seeing other people's use cases is helpful.
SPEAKER_00 (31:26):
So that's why like it's always a topic every single
week.
And then we have like tons ofsharing uh channels where people
just every day.
We have one that's called MyLife with AI.
And every day there are like 10,20 posts on it of people saying,
like, oh, Cloud Code wasterrible on this one.
Oh, Cloud Code was amazing onthis one.
(31:46):
And people just build thatcontext internally on like what
is good at today, what isbecoming good at today, etc.
etc.
So like you you need to createlike this very, very strong
connection.
SPEAKER_03 (31:57):
It's such a cool period of time.
It's like a level setting whereno matter how senior you are,
everybody's on the same learningplane, which is so, so cool.
SPEAKER_00 (32:04):
Yeah, and that goes back to what my friend was
telling me.
My knowledge, I need to justre-relearn.
SPEAKER_03 (32:12):
Relearn.
That's a good way of putting it.
We're all relearning, rewiringourselves.
SPEAKER_00 (32:16):
Rewiring, yeah.
SPEAKER_03 (32:17):
And Michelle, you have a technical background.
You're a technical founder.
What was it like to build ago-to-market engine as a
technical founder?
SPEAKER_00 (32:27):
Very good question.
Um so the first thing is I'm notalone in this adventure.
My co-founder is uh is uh alittle bit more on the on the
marketing team, on the marketingside.
It was actually the in a way itwas the first devrail of
Airbite.
So we're working together onlike technical papers and
technical articles, butotherwise he was doing a lot of
(32:48):
the heavy lifting when it comesto writing content.
I go for like I I understandpretty well like the the
psychology of users.
So we started with a very, verystrong uh bottom-up motion.
And this is a place like even ifI don't have experience like
(33:11):
building a go-to-market engine,we did pretty well at building
that bottom-up motion.
The place where I needed alittle bit more support uh was
on like how do we do the thetop-down, how do we go to toward
enterprise.
SPEAKER_03 (33:26):
How did you get that support?
SPEAKER_00 (33:28):
So I I actually hired um uh a CPO that had been
working solely on enterprise umum companies.
But I took someone that is notjust a product person, but
really someone that has a verygood, that has a lot of breast
in terms of like both product,but also like how do you
(33:50):
actually create this engine,this enterprise engine.
So to me that was the the firststep there.
Uh it started in um 2024, Ibelieve.
Yeah, that was January 2024.
And from there we started tolike bit by bit build the
enterprise engine, startingsmall at first, because you need
to learn.
Yeah, and yeah, when uh went allin there uh at the beginning of
(34:13):
the year, because yeah, 2024 isreally when we we launched the
the enterprise product and very,very quickly picked up.
So we had to uh we had to um toexpand there.
We did do a a few a few mistakesalong the way, which is selling
to enterprise takes time.
(34:33):
And when you're used to likebottom-up motion where
everything goes super fast,every deal gets closed in like a
week, two weeks, one month max,suddenly you are like in this
longer sales cycle, a lot morestakeholder, like hiring people,
you need to rent them, etc.
etc.
So what I've learned here waslike you have to be a lot more
proactive in thinking abouthiring over there.
(34:55):
So that that was that was a biglearning for me.
SPEAKER_03 (34:59):
And the ramp time proactive, do it earlier.
SPEAKER_00 (35:01):
Would that be the distilled lesson for everyone?
Yeah, while still being likebecause it takes like six to
nine months to actually deliver,you want to also edge your bet a
little bit.
But that's the that's the idea.
Like it's not gonna happen fromone week to the next.
It's gonna take a lot more time.
SPEAKER_03 (35:21):
It's a a great piece of advice for anyone listening
to.
Because most of the time,companies are wanting to go a
little bit more enterprise, andit's challenging to cross that
chasm unless you'reintentionally planning for it,
which sounds like a big lessonon your side.
SPEAKER_00 (35:37):
And there are more physical limits.
When you go bottom up, there isa lot of things that you
automate.
Like you have Saleserve, youhave like a very automated uh
sales cycle, but when you go toto enterprise, well, it's a lot
of like human time.
Um yeah.
SPEAKER_03 (35:53):
And did you feel like you went enterprise?
Why did you go enterprise?
I guess.
Were you seeing signals or wasit that you wanted to go
enterprise?
SPEAKER_00 (36:01):
No, we we have about uh 20% of our um of the Fortune
500 that are using AirBuy today,we're working like with like
very, very big media company orbanks.
And I could feel like the thelack of maturity of the team on
(36:21):
like how do we how do we sell tothat audience, how do we sell
the the product, and also whatis missing in the product.
Like when you're selling to uhdata teams, well they have their
own requirements, but when youstart selling like across
different uh business units oracross different teams, like
there are suddenly a lot morethings that you need to be
(36:43):
adding to the product that arenot directly tied to the value
that you provide, but that areactually tied to how this uh
company actually buys softwareand actually uh leverage
software.
So that was it's it's both onthe go-to-market side, but it's
very, very tied to the to theproduct.
SPEAKER_03 (37:00):
Okay.
Michelle, you raised your SeriesA two months after your seed
round.
Take us through that process.
SPEAKER_00 (37:09):
Yeah.
So we started the like raisingour seed round in November 2020.
All was finished in uh inJanuary.
By the way, we had to uh delaythe announcement because we're
trying to buy the domain.
And we didn't want to pay thepremium of uh being funded.
SPEAKER_03 (37:29):
So you had your round, you just didn't have the
domain.
Did you have a website on adifferent domain?
SPEAKER_00 (37:34):
Yes, we did.
SPEAKER_03 (37:35):
Okay, but you're trying to get the main domain
for the announcement.
How did you get it?
Did you have to just work outthe money or did you go
negotiate?
SPEAKER_00 (37:41):
No, but we it we we had it for like a a a good uh a
good price.
Okay.
Perfect.
You got it.
Secure lots of tractation herein front.
Um and the thing that happenedafter we actually announced our
series uh our seed, this is thefirst time we had put our slides
(38:02):
um live.
And it really created a lot andlot of traction on the open
source product.
Like people, because it wasreally solving a very, very,
very painful problem for thataudience.
And our numbers went likethrough the roof between like
(38:23):
January and May.
And that's also when we startedto build the engine to make sure
that contributors could be alsoinvolved in the project.
Before it was just us buildingbecause there was a lot of
foundational work that needed todo.
But we opened up the repo forexternal contribution.
I don't know, it was aroundMarch or beginning of April, and
(38:46):
it picked up really fast.
And I think at that point, whenyou see an industry that is
moving so fast, like data, uh atthe time it was not even AI, it
was just data, you see thatboom, suddenly we are present in
like 5,000.
Um I don't think it was a littlebit less.
Um, it was maybe a thousand uhdifferent companies after just
(39:08):
releasing the the repo for likea few months.
That creates a lot of uh ofattention.
And I think it's a veryinnovative way of like solving
the problem of how do you movedata around.
So that's uh that was uh I thinkthat I think they did a good
move, like going for like and wedid a good move on uh on on
(39:28):
raising the series A here, andit also allowed us to just
invest more into growing thecommunity.
SPEAKER_03 (39:34):
So are there any drawbacks to doing that?
Yes.
A lot of companies areevaluating timeline, and we
speak to many companies andadvise them around timelines,
and two months is very quick.
SPEAKER_00 (39:47):
Yeah.
SPEAKER_03 (39:47):
Now, what are the drawbacks or the pros to doing
so?
SPEAKER_00 (39:50):
Well, the one that is very simple is when you
release um an open sourceproject, you don't have you
don't get paid for it.
Yeah.
So the drawback is that uhsuddenly it puts you on uh on um
(40:13):
the expectations are high.
That's that I would say that'spretty much it.
But at the same time, you know,when we when we raise the series
A, and even when we raise theseed, we chatted with these
investors, and all the time wewere pick we were picking the
ones that had a very deepunderstanding of what it what it
(40:36):
means to build open source.
What it does it means to buildan open source company.
Because you don't do open sourcefor the sake of doing open
source, you do it because youhave a strategy.
And ours was very strongbottom-up awareness, building a
standard, and those can take alittle bit of time.
You know, you can look at youcan look at Elastic, you can
look at Ashi Corp, etc.
etc.
(40:56):
Like all of these, like youcreate a very strong base, yeah,
and then you figure out like allthe different basically your
real product market fit.
Um, and so I would say like notlike that's a risk of drawback.
We did not have it because wehad a a very uh knowledgeable um
uh investor on that front.
SPEAKER_03 (41:17):
Got it.
So it sounds like a learning foranyone thinking about this kind
of strategy or even just overallwith the alignment around
expertise with your investor.
SPEAKER_00 (41:27):
Exactly.
Okay.
The partner you're working with,well, yeah, they're gonna be
here for a very long time.
You better be very aligned withthem on like what you want to do
and also like their tolerancefor yes, things don't always go
right.
SPEAKER_03 (41:44):
And how do you evaluate that from the founder
seat?
Because naturally we evaluate itall the time from the other
side.
SPEAKER_00 (41:49):
Yeah.
Um, well, like always, when youin a way you you recruit
someone, yeah, back channels isthe best way.
So you talk to other companies,you you you search for the
company where it went well, theone that where it didn't go
well, and create a relationshipwith the with the people that
have been working there and andsee what they have to say.
SPEAKER_01 (42:09):
So excellent.
SPEAKER_00 (42:11):
And also you see, you know, you you also see like
during the during the thefundraising process, like is how
much are they um evolving yourthinking?
Uh, you know, when we raisedwith Axel.
SPEAKER_01 (42:27):
Yeah.
SPEAKER_00 (42:28):
Like I remember spending like two or three hours
with uh with Amit at the time,and he asked questions that in a
way helped us improve how wewere thinking about the the
future, the positioning ofAirbag, and what to do.
So there was already some verystrong value on like working
with uh with him or working withuh with Shetan at benchmark.
(42:50):
It's like they they help youthink.
And yes, they have theiropinion, I have my opinions, but
at the end of the day, like arethey allowing you to see places
what you don't know about?
SPEAKER_01 (43:04):
Right.
SPEAKER_00 (43:04):
And if so, I think that's a that can become a great
partnership.
SPEAKER_03 (43:08):
Great advice for anyone listening, thinking about
that investor founderrelationship.
Okay, take us back to a littlebit earlier.
If we were to circle back toyour product market fit, were
there any kind of biggestchallenges to that?
I think there were some big kindof moments around that.
SPEAKER_00 (43:30):
Um yes.
Um I would say like in 2022,that's when we we started to
work on the on the cloudproduct, which by the way, if
you're an open like for opensource founders, like going from
an open source product to anactual cloud product, it is
super hard.
(43:51):
Because hosting and managingsomething when what you've done
is like providing something thatyou don't need to really host
and manage, etc.
etc., this is very, very hard.
And in 2022, we released likethe let's call it like the
private beta of uh of AirbiteCloud.
There was a ton of problems, andwhich by the way is completely
(44:14):
normal, but we underestimatedhow complex it was to build a
platform.
And because we had this big planof like how we're gonna be
monetizing airbite, etc.
etc., I would say we we hired alittle bit ahead.
And that to me was uh was amistake because it also creates
a lot of noise internally.
(44:35):
It like disrupts the productteam, it disrupts engineering,
it creates like a lot of noisearound like building the best
product.
And that to me was uh I don'tknow, I would say was a bad
decision.
Uh we we had to course correct,but I would say it's like
(44:56):
especially when you're startingsomething new, just start small,
expand rather than go gobottom-up in terms of how you're
building your your your companyand your organization rather
rather than top-down.
There is a moment when you cando top-down when you have like a
lot more predictability, but atthe beginning, it's uh it's a
mistake.
SPEAKER_03 (45:14):
So bottom up.
SPEAKER_00 (45:15):
At least for us it was a mistake.
SPEAKER_03 (45:16):
Yeah, well, it sounds like in everything that
you did, you were always lookingat signals.
Like I know I've heard you saythat open source allowed you to
have signal density.
And through the communityaspect, you're talking about
using that as signals andfeedback too, and same with this
bottoms-up approach.
SPEAKER_00 (45:31):
Yeah, yeah.
I'm a I'm a very bottom-upperson on that front.
But at some point, yes, when Isee it's always the same thing,
like we all build we're buildingan engine.
So we need to figure out what isthe the MVP of that engine.
unknown (45:46):
Yeah.
SPEAKER_00 (45:46):
And to do that, you need to find people that are
extremely driven, that are okaywith uncertainty.
But the moment you start gettinglike an initial version that is
working, that's when you canstart like putting more uh like
more thought into what it shouldactually look like.
But first you need to validatesomething.
Definitely.
SPEAKER_03 (46:05):
Well, this has been fantastic, Michelle.
Really, really appreciate thetime.
A couple last questions.
You know, you have learnedimmensely throughout the
journey, but are there any booksthat you've particularly been
influenced by throughout yourcareer and life?
SPEAKER_00 (46:20):
Yeah, you see I don't know if you remember, but
I said giving a window into howthings are working to the
outside world.
Yeah, I did not invent that.
It's like when I was I think itwas in 2014, I was just um, or
2013, I was just starting to uhto manage my my first team, my
first team at the time.
(46:41):
And my CTO gave me this bookfrom um it's called Um High
Output Management.
I think it's now it's astandard.
Uh and it really, you know, whenyou go from being an IC to
starting to manage people, it'svery hard to find like the right
feedback clue for like, are youdoing a good job or not?
(47:02):
Like what does it mean thatyou're doing a good job?
And also how do you build teamas systems?
And I think that book was justtransformational for me because
I like good theory, yeah, andthat theory was very very
strong, and like how you create,how you how you build, how you
build the system, how youmonitor these systems, and um
(47:25):
and uh how you take pride of thework when you're not the one
always doing the work yourself.
SPEAKER_03 (47:31):
Yeah.
So great, fantastic.
Well, that will be in the shownotes for anyone curious to take
read and pay it for it a littlebit.
Where can people follow alongyour journey in airbytes?
SPEAKER_00 (47:40):
Uh well, the I would say the entry point is always
gonna be airbite.com.
SPEAKER_03 (47:45):
There we go.
SPEAKER_00 (47:46):
I'm on I'm on LinkedIn, I try to post as much
as I can.
SPEAKER_03 (47:49):
Yeah.
SPEAKER_00 (47:50):
Uh content marketing.
There we go.
You're very good at it.
And giving and giving a windowto uh to the to the people uh on
what we're doing.
And and yeah, and after that,like you can go on Slack, on our
GitHub repository, and just orjust try the product.
SPEAKER_03 (48:06):
There we go.
There's many ways.
Many ways.
SPEAKER_00 (48:08):
Point of entry is going to be the website.
Perfect.
SPEAKER_03 (48:11):
The website itself.
Well, Michelle, this has beenfabulous.
Really appreciate the time andyou sharing your journey and
insights.
SPEAKER_00 (48:15):
Yeah, thank you for having me.
It was a great conversation.
SPEAKER_03 (48:18):
Absolutely.
People are willing to put time into the project and
the product that we arebuilding.
How do you actuallycommercialize it?
It's a different story.
And to me, that's what PMFactually is, where everything
goes super fast, every deal getsclosed in like a week, two
weeks, one month max.
SPEAKER_03 (00:15):
You launched in 2020.
Now you're valued at over abillion dollars.
Michelle didn't play the typicalfast launch playbook.
He went open source first,community first, and GitHub
first.
And it ignited one of thefastest bottoms-up adoption
curves in modern datainfrastructure.
Today, AirPyte is valued at overa billion dollars, powering data
movements for thousands ofteams, including over 20% of the
(00:37):
Fortune 500.
And they got there in a reallyinteresting way.
They built in public, compoundedthrough community, and turned
contribution into a distributionmode.
In this conversation, we breakdown the stories and lessons
behind all of this growth,including a really important
lesson on separating productmarket fit from project market
fit.
All right, let's get into it.
(01:09):
Michelle, welcome to thepodcast.
SPEAKER_00 (01:11):
Thank you for having me.
Great to be there.
SPEAKER_03 (01:13):
It is a pleasure.
And it hasn't been long since wesaw each other earlier this
week, in fact.
So it's great to see you again.
SPEAKER_00 (01:19):
Yeah, no, that was a good, a good event.
Like that was the Tech Crunchone that was very solid.
SPEAKER_03 (01:25):
Yeah, that was great.
And your your uh session washighly attended.
Sounded fantastic.
Exciting to pick your brain alittle bit more intimately than
at the event itself.
And you launched in 2020.
Now you're valued at over abillion dollars.
Take us back.
We want to know the how behindthis type of growth.
SPEAKER_01 (01:43):
Yeah.
SPEAKER_03 (01:44):
And before we even get started from the beginning,
a lot of our audience aren'tengineers.
A lot of operators, a lot offounders.
Give us a high level of Airbite.
What does Airbite do?
Kind of the two-liner foreveryone listening.
SPEAKER_00 (01:57):
Yeah.
So AirBite is an open datamovement platform, meaning that
we can take any pieces of dataacross any system and we can
deliver it into a place where itwill deliver value.
So a very strong use case isgoing to be everything related
to analytics.
How do you go across yourcompany, look at all the
services that you have, all thedata sources that you have, all
(02:19):
the silos that you have, and howdo you make it seamless to move
that data into warehouse so thatyour analytics team can actually
extract insight from it and makedecisions from it.
And that's really how westarted.
There is a ton of use case whenit comes to moving data.
You know, we're talking aboutagents these days, is like how
do you get the data into agents?
So that's very much what thevery high-level value of Airbite
(02:44):
is.
SPEAKER_03 (02:44):
Super helpful.
And now let's go back to thebeginning.
You launched on GitHub.
Why open source as opposed to atraditional product launch?
SPEAKER_00 (02:53):
Yeah.
So when you're thinking about,let's take the analytics use
case as an example.
You go from like the outcome youwant to drive, which is I want
to be able to understand mybusiness.
The first thing you think aboutis okay, I will need to have
dashboards, I will need to havea team, I will need to have a
warehouse.
(03:14):
And the moment you have thesetwo, what you realize is that
you also need the data,obviously.
SPEAKER_02 (03:18):
Yeah.
SPEAKER_00 (03:19):
And this is a very organic um behavior from people,
which is it's not thoughtthrough so much as a strategy,
but more as an enabler.
So they're gonna go bit by bitthinking, oh, I need this
particular silo, I need thisparticular silo.
And it is very hard to actuallythink about the pain that it
(03:43):
will be if you build ityourself, or it will be very
hard also to find platforms thatcan support every single silos
that you have.
And for us, when we did opensource, what we wanted is to go
and talk to the team that arebuilding all these different
connectors.
So when you're an engineer andyou're being asked, oh, I need
(04:06):
Stripe data to be in thewarehouse, the first reflex that
an engineer will have is goonline, check how do I move data
from Stripe, Salesforce, HubSootor you name it, into my
warehouse.
And we wanted to catch thesepeople exactly at that time.
We wanted to provide them valuethe moment they have that little
(04:27):
painful script that they have towrite and give them something.
So open source at that point isgenerally the best solution
because I mean I'm an engineer,I'm a little bit lazy when it
comes to if I can avoid buildingsomething, I will.
SPEAKER_02 (04:42):
Yeah, fair.
SPEAKER_00 (04:42):
And open source is generally the solution for that,
and that's really why we wentfor like open source.
The other reason is there's aninfinity of places where data
can be.
So it is impossible for a singlecompany to make a product that
will address all the long tailsof data connectors.
What we need is, and what thecommunity needs is like, in a
(05:05):
way, all working together in agoal of like addressing all
these use cases.
And that's why open source forus was a solution.
Like you, you know, you canthink about the Linux kernel.
Well, all the drivers are beingbuilt either by the community,
either by by vendors, but theLinux project is not building
all these drivers.
(05:26):
They are asking the community tobuild those, and that's how you
just get to the best uh uhproduct on the market.
SPEAKER_03 (05:34):
And it feels like we're seeing more and more
companies open source.
Do you feel that also?
SPEAKER_00 (05:39):
Yes.
Um, yes, and I think it'sbecause the technology,
especially this, you know, opensource is very, very present in
AI, for example, because thereis almost like a complete stop
of the old world versus the newworld.
Like everything has to bereinvented.
And people who are makingdecisions today have to catch up
(06:03):
on a lot of context.
So, what they do is actuallythey go talk to their team and
ask them we I we need to createan agent for this particular use
case.
What technology should we beusing?
And open source generally worksreally well with technical
profiles.
And I think that's one of thereasons.
There are also a lot of thingsaround sovereignty and control
(06:25):
that comes with open source andalso future proofing because you
can always update the projectyourself if you want to.
And to me, that's a directionthat we're seeing.
And having a community thatbacks a project just you cannot
beat that velocity.
SPEAKER_03 (06:42):
Yeah, so true.
So true.
And okay, so you launched in2020.
When you uploaded the repo, didyou know that it would take off
the way that it did?
SPEAKER_00 (06:53):
No, we didn't know.
We are so in the story ofHairbyte, like Airbyte started
really just two months beforeCOVID really hit the world.
SPEAKER_03 (07:02):
What a time to start.
Yeah.
SPEAKER_00 (07:04):
And we had an initial product at the time,
which was also related to dataintegration, but more geared
toward marketing teams.
And what happened with COVID isboom, all the marketing team got
frozen, laid off, etc.
etc.
Because company had to figureout, okay, what does the world
look like now?
And you know, as a founder, youput your life into uh a company,
(07:27):
into building a product, and youdon't want to be a vitamin that
I like to joke about that is notgoing to survive a global
pandemic.
So what we did is we actuallywent back to the drawing board.
And in July, like during theperiod of like March to July, we
were building prototypes, etc.
etc.
But but we're also talking a lotwith the audience that we wanted
(07:49):
to build a product for, whichwas data people.
And all these people, they werealways having a solution that
they would buy, a solution thatthey will build, another
solution that they would build,another solution that they would
buy.
So it was like a collection oftools everywhere just to move
data.
And what we've done is justkeeping in touch with all these
(08:10):
people and keeping them in theloop of what we were building,
what product.
SPEAKER_03 (08:15):
What are you doing that?
How are you keeping them in theloop?
SPEAKER_00 (08:18):
So at the time during COVID, everybody, I think
a lot of people were veryavailable on LinkedIn.
SPEAKER_01 (08:23):
Yeah.
SPEAKER_00 (08:23):
So we're very, very active on LinkedIn.
So we were always trying to talkto the right people, going on a
Zoom with them for like 15minutes, 30 minutes, and then we
would ask them, Do you want tobe following what we're doing?
And say yes.
And then we created the firstmailing list that we had, and
every time we had updates, wewould just say, Oh, this is what
(08:44):
we're building.
If you want to, we can give youa quick demo of what it looks
like, and you can give usfeedback.
That was before we published therepo.
And I think it was in Novemberwe actually put the um the repo
out.
And suddenly, first of all, likethis initial group of people
started to download the thesoftware, started to give us
(09:06):
like real feedback, and fromthere it just went uh in hockey
stick.
SPEAKER_03 (09:11):
Yeah, incredible.
How did you feel just seeingthat growth after you said it
yourself when you're a founder?
You put you put everything intoa company.
SPEAKER_00 (09:19):
Yeah.
It's uh I felt very good in away, which is people are willing
to put time into the project andthe product that we're building,
and yet it is super immature.
And you know, we always talkabout PMF in the the founder
founding sphere.
PMF, my definition, having seenthat, is it's when people are
(09:40):
willing to go above and beyondto make something that is not
yet mature, that is not yetworking, and they are willing to
put the effort to make it workbecause it is solving such an
intense problem for them thatthis little pain of making it
work is better than the big painof having to do it yourself.
(10:01):
Um, and yeah, it felt good.
After that, yes, I knew that thetechnology needed to become
better, but you have to launch.
SPEAKER_03 (10:08):
Yeah, yeah, exactly.
Usually, if you're at a pointwhere you feel like it's good
enough, it's too late from alaunch perspective.
SPEAKER_00 (10:15):
Exactly.
Like you want to get thefeedback as fast as possible.
You just want to build what isactually going to deliver value
for your community.
SPEAKER_03 (10:22):
A quick pause to tell you about a company you
need to know.
ZoomInfo is the go-to-marketintelligence platform built for
sales, marketing, and rugoffs.
By unifying data, workflows, andinsights into a single system,
ZoomInfo helps revenue teamsfind and engage the right
buyers, launch go-to-marketplacefaster, and drive predictable
growth.
With industry leading accuracyand depth of data, it gives your
(10:42):
team the intelligence advantageto win in competitive markets.
It's trusted by the fastestgrowing companies and has become
the category leader ingo-to-market intelligence.
Learn more at zoominfo.com.
So, how long did it take to hitPMF in your definition project
market fit?
SPEAKER_00 (11:01):
I'm actually splitting it because there are
two paths in the life ofAirbytes.
There is what I call projectmarket fit, which is we managed
to create a project that wasvery much resonating with an
audience, data engineers, dataanalysts, etc.
And they were just taking theproject and using it and
(11:22):
contributing to it.
Product market fit for me alsocomes when you start pulling the
foundation also of uhmonetization.
And this is a different storybecause it's easy to take a
product from GitHub.
How you actually commercializeit, it's a different story.
And to me, that's what PMFactually is.
So I would say open source wasproject market fit.
SPEAKER_03 (11:45):
Got it.
Okay.
Well, take us through a littlebit of the evolution then.
Because you got project marketfit.
What kind of go-to-marketdecisions did you make along the
way that helped you to get toproduct market fit from project
market fit?
SPEAKER_00 (12:00):
We were not using regular channels.
It was all about content.
I mean, at the time, contentmarketing was a thing, but I
don't think it was as uh aspopular as it has been like in
2023, 2024.
But we were just always pushingarticles, giving details about
how the what the company isdoing, what the project looks
(12:22):
like, and just getting people tobe part of our adventure.
And that created trust, thatcreated curiosity, that created
a lot of awareness.
You know, we published ourfundraising slides, for example.
So that was a way for us of likeengaging the community into what
we are doing.
So that uh that to me issomething that not a lot of
(12:43):
people have done in the past.
So it was very uh I think veryI'm I'm pretty proud that we've
done that.
It's very, very innovative.
And then, yeah, like we'vealways been very strong on
content, engaging with thecommunity.
SPEAKER_03 (12:57):
So but it is time consuming content.
So how do you think about thatas a founder?
How do you balance yourschedule?
When do you work it in?
What's your actual cadence orwas at the time for any kind of
founders or operators listeningthat are looking to up level
their content?
SPEAKER_00 (13:11):
Yeah.
It is time consuming, but youknow, if it's working and you
feel it's gonna be workingbetter than any other solution,
you just continue and you youexploit that uh that channel as
much as you can.
Um after that, we yeah, we arenow we're doing we continue to
(13:31):
do a lot of content, but we arealso a lot more like traditional
channels like ads, uh SEO, GEO,etc.
etc.
SPEAKER_03 (13:39):
So yeah.
And did you write it allyourself?
Did you hire a ghostwriter?
Like when did you actuallyphysically put kind of uh pen to
paper, if you will, or fingersto keyboard?
SPEAKER_00 (13:51):
I would say the first year and a half, it was my
co-founder and I writing.
The team also was writing.
So we really created that cutethat internal culture of let's
write something.
My VP of engineering wrote anamazing article about the pain
of building connectors that wekeep referring to, even five
years later, uh, because itreally explains the pain.
SPEAKER_03 (14:14):
Yeah.
Well, it's funny too.
Five years later, and the painis still the pain.
SPEAKER_00 (14:18):
The pain is still the pain, and uh yeah.
SPEAKER_03 (14:20):
Yeah.
Incredible.
Okay.
So you lean into content early,and that helped sounds like
create a bit of a tribe, and youhave a very strong following of
people that are passionate aboutthe product and the space and
the solution that you built.
How did you think about actuallytaking that interest generated
from your content and othermeans and turning it into more
of a community motion?
SPEAKER_00 (14:41):
Yeah.
For me, at that point, so we wealso created a Slack community
at the time.
I think today we have about25,000 people on it.
And the way we created thecommunity was in twofold.
One is we were helping thecommunity a lot.
(15:02):
We are doing a lot of supportbecause we are building the
platform.
And so every time someone had anissue, that was a product
feedback for us.
So we spent a lot of time in2020, like end of 2020,
beginning of 2020, like all of2021, and uh we we continued
after, but that was very, veryintense, a year and a half,
where we were always on Slack.
(15:24):
Every single issue that wasreported, we would just have
something shipped the day afteror like the week after.
So I think that created that'sthat was one thing that created
that helped uh building thecommunity.
And then what we did is we alsoidentified a lot of champions
within the community, likepeople that wanted to help other
people.
(15:45):
And yeah, we really engaged withthem.
We actually hired one of thefirst community managers that we
that we've had at uh at Airbite,is someone that we actually
brought from the from thecommunity that was he started to
build an airflow connector, likean airflow integration, and say,
oh man, that's amazing.
And we didn't ask him anything,and at some point we asked him,
(16:05):
like, do you want to uh to do ityour full-time job, like to
engage with the community, writecontent, etc.
etc.
And they're like, Yeah, let's doit.
SPEAKER_02 (16:12):
So it's amazing.
SPEAKER_00 (16:13):
That to me is like the community engagement is is
absolutely key.
That's how you create thattribe, that's how you create
that snowball effect.
It's it's not something that youput on, you say, I'm building
community, and it's gonna happenby itself.
No, it is something that has tobe worked on, and you have to be
intentional about what you wantto do uh with the community.
SPEAKER_03 (16:34):
If you were to look back now with the benefit of
hindsight, it sounds likecommunity and content are two
pillars that helped with yourgo-to-market motion.
Yep.
Is that correct?
And also are there other pillarsthat you'd say were really
pivotal in your growth?
SPEAKER_00 (16:49):
Um we did a lot of events, actually, very
specialized events around uhdata, whether they were open
source events or like Snowflakeor Databricks events, it's just
always getting where the peoplewere.
And that was something thatworked pretty well.
(17:11):
It allowed us to get a lot ofpeople, new people interested,
or just to engage in real lifewith people.
Yeah um yeah.
I would say, and here's reallywhat happened between 2020 and
2023.
After that, we had we added afew other things on top of uh
like how we engage with thecommunity, etc.
(17:33):
etc.
But that to me was very verymuch like the three pillar of
what we've done.
It's like giving a window intothe company to people, giving in
a window into how the theengineering team is building,
giving a window into everythingwe're doing.
SPEAKER_03 (17:49):
And do you still operate that way?
SPEAKER_00 (17:51):
Uh a little bit less.
Um but we continue to have thatconstant engagement with uh with
people.
Like, you know, when the thegreat thing that when you have a
community like that is someonein whether it's a customer,
whether it's uh it's a user, isgoing to ask you a question or a
(18:15):
feature, and then you're it'sgonna go into your head and say,
okay, is that really useful?
Or is it just for that person?
And so what you do is you go inyour on your community and you
you just post a very simplequestion, like is that something
that resonates with you?
And in 30 minutes, you have likehundred people that are
(18:37):
replying, yes, no, yes, but inthat way.
So it really accelerates how youdo product discovery, how you do
uh product development.
So that's uh that's more likehow we've changed a few things
along the way.
It's like we're we're leveragingthe the community a lot more for
like what new features we shouldbe building rather than really
the the the core valueproposition of the ambite.
SPEAKER_03 (19:00):
Right.
It's uh a feedback loop, yeah,essentially.
SPEAKER_00 (19:03):
Yeah.
SPEAKER_03 (19:03):
Great, and a very, very rapid one too.
SPEAKER_00 (19:06):
Very rapid one.
SPEAKER_03 (19:07):
So content, community events, pillars that
you did incredibly well to reachthe point you are now.
There's always the other side ofthe story of you know, what were
the the areas that didn't quitehit as well or almost the
near-death experiences along theway that every startup goes
through?
SPEAKER_00 (19:22):
Yeah, so as I said, like the beginning of the of how
we are engaging with thecommunity was very a lot of
support, like helping them besuccessful with the product.
And there was this moment whereeven in our in how we were
working, our community becamevery much of a like support
channel rather than likebuilding a uh a community that
(19:49):
was just helping each other.
Um, and that to me was uh issomething that we could have
been more intentional at thebeginning around how do we um
how do we get to like communitymembers helping each other,
community members like meetingeach other outside of just like
rather than becoming a very muchlike support-oriented um uh
(20:13):
community.
And the thing is, once thishabit is taken, it's very hard
to shift uh into a differentdirection.
I think we succeeded, but ittook us a lot of time.
We should have been moreproactive thinking about okay,
the community is amazing, butwhat is the future?
Like, how do we make it morevibrant, more um yeah.
(20:33):
How do we create a community ofprofessional that work in data
and that are just gonna learnfrom each other and not just
from us?
SPEAKER_03 (20:41):
Mm-hmm.
Yeah, it completely makes sense.
It's kind of the the the tellall tales, the tell-old tale
story of community is a lotharder in practice, and it does
require some really deepintentionality around fostering
it.
SPEAKER_00 (20:55):
It does, it does.
SPEAKER_03 (20:56):
Yeah, and what does that team kind of composition
look like right now at Airbite?
SPEAKER_00 (21:01):
Um so we have we have a we have a DevRel person,
and this this person is more umfocused on the like the content
strategy geared, oops, gearedtoward the community.
And we have a community manager,meaning someone that just
engages, identifies champion,uh, gives them access to um
(21:26):
early features, etc.
etc.
And we also have people um ininternally we call them like
customer engineering, wheretheir focus is to make sure that
every product feedback aroundconnectors is being funneled
through the team to make surethat our connectors keep getting
better and better and better.
(21:47):
So this is more like for thecontributors of the platform.
So we really have a differencebetween like the users of the
platform and the contributors ofthe platform, and we handle
these two groups differently.
SPEAKER_03 (21:58):
Gotcha, gotcha.
Okay.
What are some other areas thatyou know along the journey,
again, reflecting back, havejust been some of the most
pivotal things that maybe youdon't you don't see or talk
about as much?
SPEAKER_00 (22:11):
Um I think that was the the realization of why are
so many companies using airbite.
Is it just connectors or is itsomething else?
And connectors is a is levelone, but there is a a second
level to it, and it took us alittle by a little bit of time
(22:33):
to figure it out, is people werealso using data, uh airbite,
because there was so much redtape around the data that they
had internally, that having aplatform that they fully control
that runs within theirinfrastructure, it's a byproduct
(22:55):
of open source.
And we did not realize uh Iwould say like fast enough that
that was one of the key reasonswhy so many teams were adopting
airbytes.
So, you know, when we started toto do the airbyte monetization,
we said, okay, we're gonnafollow the we're gonna skip the
step of doing support for peoplethat are deploying airbytes, and
instead we're gonna go directlyto a cloud product.
(23:17):
And very quickly we realized,yes, cloud is getting traction,
but we are not able to convertevery person that is using
airbite to using airbite cloud.
And at that point, we just wentback to the drawing board,
started to talk to them, andthat's when we discovered that
in that case, like productmarket fits was not just
connector.
It was the fact that these pipeswere under their control.
(23:40):
And that was a big, a big thing,and I would say we we wasted a
little bit of time on trying tobuild something fully cloud when
what people needed was controland sovereignty.
SPEAKER_03 (23:53):
Got it.
Okay.
SPEAKER_00 (23:55):
More like, you know, when you're searching for PMF,
it's not a straight.
SPEAKER_03 (24:00):
Never linear, never linear, no.
And what are you most excitedabout thinking now forward?
SPEAKER_00 (24:06):
Yeah.
Well, you know, every every timeI hear about how do I make I
mean to me like the AI wave thatis happening right now is just
one of the most exciting thingsfor me and for the for the
company.
Like analytics is very much acore part of what we're doing,
but we're getting so much pullinto different types of data
(24:30):
access.
And that is something that we'retoday encoding into the platform
and into our connectors.
It's not just humans consumingdata today.
Yeah, it's agents that candiscover what's available,
discover what it looks like, andmake decisions.
So, yes, the technology is notyet completely mature on either
side, whether it's airbite,whether it's like agency
(24:50):
platform, etc.
etc.
But you can see how fast it'smoving, and I think it's very
energizing, especially in theinfrastructure world, to see
that that energy being uh beinginjected.
So that yeah, I I I talk aboutit all the time.
SPEAKER_03 (25:06):
So Yeah, no, that's fantastic.
And I mean you mentioned that atthe very beginning around how
now it's agents consuming thistype of data.
How does that transition in theoverall industry?
What does anyone need to knowabout what this transition
actually means?
SPEAKER_00 (25:22):
Yeah.
You need to you need to forgetabout a lot of your existing
patterns.
You know, I was chatting with uhwith a CTO uh last week, and he
told me very bluntly, I don'tknow, maybe he was trying to uh
(25:42):
to be a little bit uh uhprovocative here, but he said he
told me, Michelle, all thetechnical knowledge I had
stopped two years ago.
I had to fully reinvent myselfand reinvent my team.
Uh so yes, some things are stilltransferable, but your default
should always be thinking abouthow do I build in that new
(26:05):
world?
Is there a solution?
No.
Okay, maybe I go back and usethe techniques of the of the the
older world.
But that's really what what I'mseeing is people have to rethink
how they are doing their job.
Because one thing that ishappening in Teams is a lot of
people are using AI today toremove from their play the thing
(26:28):
that they don't like doing.
That's very easy.
Like people have a very stronguh willingness to stop doing the
things they hate doing.
So for that, like AI is is isamazing.
Like, you know, if you're anengineer like writing unit
tests, writing integrationtests, that's great, but that's
just level one.
The moment you actually startchanging your mindset is when
(26:49):
you're looking at the things youlike doing and how can you
leverage AI for those.
But those are hard because thethings you like doing are the
things also that can bring you alot of energy in your in your
day-to-day.
And those are the things thatpeople should really be focusing
on.
On okay, this thing that I'mdoing every day, I love doing
it, but can I do it using AI,using an agent?
(27:12):
Can I ask my engineering team tobuild an agent to solve that
particular problem?
Is there an AI product thatexists that can do it and
removes that from my plate?
And then I can focus on morethings and I can become faster.
But to me, it's really aboutreinventing um reinventing it.
For data, the way you accessdata is very different.
SPEAKER_01 (27:33):
Yeah.
SPEAKER_00 (27:34):
Um but just having a warehouse doesn't cut it.
Like you need to have like anagent does live processing, it
needs to have like little piecesof data here and there.
You need to provide access tothe agent in a different way.
So that's and that's what that'swhat we've been building.
SPEAKER_03 (27:54):
Yeah, absolutely.
How did that change your productroadmap overall?
Did you have like this crazymoment in a way where it was
like the realization that youentirely have to pivot?
Or is it gradual?
SPEAKER_00 (28:07):
I I wouldn't say it's a it's a pivot because it's
more like a an extension, butalso sometimes we like to talk
about replatforming, which iswe've we've built the plat the
platform for like a specific usecase in a specific course, but
there are new ones that arecoming that are going to pick up
massively over the next fewyears.
(28:28):
And we need to be thinking abouttaking all the learnings that
we've had here and how do wethink about replatforming it to
just have a larger breadth ofuse case.
So that's that's more how howwe're thinking about it.
Uh I don't know, I would say2024 is when we even even before
like summer 2023 is is when westarted to like tippito into it.
(28:50):
But 2025 is a moment where we wewent all in on that.
So we still have the theanalytics product, it's a it's
an amazing product, but we arereally building on top of that,
like leveraging part of it, butalso rebuilding a platform that
allows agents to uh to interactwith data.
So it's pretty pretty cool.
SPEAKER_03 (29:12):
Super cool.
And you're hitting the groundrunning, tons of growth, you're
hiring lots of folks on theteam.
Like, how do you think aboutdeveloping that team to take it
to the next stage of growth?
SPEAKER_00 (29:22):
Oh, you have to be hammering using AI every single
day, every single or-ens that Ido every Wednesday morning.
It's about putting the spotlighton new uh new way of leveraging
AI.
And not just the layer one,which is do the thing you don't
like doing.
It's really about how are peoplebuilding things that change
(29:46):
their, actually change the thedefinition of their job.
So well, if it's if it's onsales, it's gonna be around like
how do they do uh like discoveryof account, it's gonna be how
they connect um uh Likedifferent news together, how do
they connect to like pastconversation that's happened on
support?
(30:07):
So it's really about likeaggregating all this information
in one single place and havelike all the context available
to them at the right time.
On engineering, well, we talkedenough about engineering and how
agents are transforming thelives of engineers, and that's
what we've been doing atAirbytes for the yeah, for the
for the past year.
SPEAKER_03 (30:26):
Yeah.
And so it sounds like youdisseminate this information
internally.
You said weekly.
SPEAKER_00 (30:31):
Weekly.
SPEAKER_03 (30:32):
What does that look like?
Everyone's on a team callweekly, or how are you
spotlighting people?
SPEAKER_00 (30:36):
Yeah, so the whole company we'd spend like 30
minutes together and we go overlike some updates, but then we
always have like onepresentation that is just about
AI.
And myself, like I generallystart the whole hand and I
always have a few slides aroundlike the wins of the week.
SPEAKER_01 (30:52):
Yeah.
SPEAKER_00 (30:52):
And well, we have a channel on Slack where people
just write their wins and I justgonna pick one or two.
And that's why I mean likepulling the spotlights on
specific individuals that havedone something innovative with
it.
And I think that creates like agood dynamic.
Like people want to be on thewin slide, etc.
etc.
So it creates a little bit oflike internal competition.
SPEAKER_03 (31:14):
Yeah, internal competition and also ideation.
Exactly.
I find sometimes the biggestblocker is the inspiration and
ideation around like what can Iactually do with AI?
SPEAKER_00 (31:23):
Exactly.
SPEAKER_03 (31:23):
So seeing other people's use cases is helpful.
SPEAKER_00 (31:26):
So that's why like it's always a topic every single
week.
And then we have like tons ofsharing uh channels where people
just every day.
We have one that's called MyLife with AI.
And every day there are like 10,20 posts on it of people saying,
like, oh, Cloud Code wasterrible on this one.
Oh, Cloud Code was amazing onthis one.
(31:46):
And people just build thatcontext internally on like what
is good at today, what isbecoming good at today, etc.
etc.
So like you you need to createlike this very, very strong
connection.
SPEAKER_03 (31:57):
It's such a cool period of time.
It's like a level setting whereno matter how senior you are,
everybody's on the same learningplane, which is so, so cool.
SPEAKER_00 (32:04):
Yeah, and that goes back to what my friend was
telling me.
My knowledge, I need to justre-relearn.
SPEAKER_03 (32:12):
Relearn.
That's a good way of putting it.
We're all relearning, rewiringourselves.
SPEAKER_00 (32:16):
Rewiring, yeah.
SPEAKER_03 (32:17):
And Michelle, you have a technical background.
You're a technical founder.
What was it like to build ago-to-market engine as a
technical founder?
SPEAKER_00 (32:27):
Very good question.
Um so the first thing is I'm notalone in this adventure.
My co-founder is uh is uh alittle bit more on the on the
marketing team, on the marketingside.
It was actually the in a way itwas the first devrail of
Airbite.
So we're working together onlike technical papers and
technical articles, butotherwise he was doing a lot of
(32:48):
the heavy lifting when it comesto writing content.
I go for like I I understandpretty well like the the
psychology of users.
So we started with a very, verystrong uh bottom-up motion.
And this is a place like even ifI don't have experience like
(33:11):
building a go-to-market engine,we did pretty well at building
that bottom-up motion.
The place where I needed alittle bit more support uh was
on like how do we do the thetop-down, how do we go to toward
enterprise.
SPEAKER_03 (33:26):
How did you get that support?
SPEAKER_00 (33:28):
So I I actually hired um uh a CPO that had been
working solely on enterprise umum companies.
But I took someone that is notjust a product person, but
really someone that has a verygood, that has a lot of breast
in terms of like both product,but also like how do you
(33:50):
actually create this engine,this enterprise engine.
So to me that was the the firststep there.
Uh it started in um 2024, Ibelieve.
Yeah, that was January 2024.
And from there we started tolike bit by bit build the
enterprise engine, startingsmall at first, because you need
to learn.
Yeah, and yeah, when uh went allin there uh at the beginning of
(34:13):
the year, because yeah, 2024 isreally when we we launched the
the enterprise product and very,very quickly picked up.
So we had to uh we had to um toexpand there.
We did do a a few a few mistakesalong the way, which is selling
to enterprise takes time.
(34:33):
And when you're used to likebottom-up motion where
everything goes super fast,every deal gets closed in like a
week, two weeks, one month max,suddenly you are like in this
longer sales cycle, a lot morestakeholder, like hiring people,
you need to rent them, etc.
etc.
So what I've learned here waslike you have to be a lot more
proactive in thinking abouthiring over there.
(34:55):
So that that was that was a biglearning for me.
SPEAKER_03 (34:59):
And the ramp time proactive, do it earlier.
SPEAKER_00 (35:01):
Would that be the distilled lesson for everyone?
Yeah, while still being likebecause it takes like six to
nine months to actually deliver,you want to also edge your bet a
little bit.
But that's the that's the idea.
Like it's not gonna happen fromone week to the next.
It's gonna take a lot more time.
SPEAKER_03 (35:21):
It's a a great piece of advice for anyone listening
to.
Because most of the time,companies are wanting to go a
little bit more enterprise, andit's challenging to cross that
chasm unless you'reintentionally planning for it,
which sounds like a big lessonon your side.
SPEAKER_00 (35:37):
And there are more physical limits.
When you go bottom up, there isa lot of things that you
automate.
Like you have Saleserve, youhave like a very automated uh
sales cycle, but when you go toto enterprise, well, it's a lot
of like human time.
Um yeah.
SPEAKER_03 (35:53):
And did you feel like you went enterprise?
Why did you go enterprise?
I guess.
Were you seeing signals or wasit that you wanted to go
enterprise?
SPEAKER_00 (36:01):
No, we we have about uh 20% of our um of the Fortune
500 that are using AirBuy today,we're working like with like
very, very big media company orbanks.
And I could feel like the thelack of maturity of the team on
(36:21):
like how do we how do we sell tothat audience, how do we sell
the the product, and also whatis missing in the product.
Like when you're selling to uhdata teams, well they have their
own requirements, but when youstart selling like across
different uh business units oracross different teams, like
there are suddenly a lot morethings that you need to be
(36:43):
adding to the product that arenot directly tied to the value
that you provide, but that areactually tied to how this uh
company actually buys softwareand actually uh leverage
software.
So that was it's it's both onthe go-to-market side, but it's
very, very tied to the to theproduct.
SPEAKER_03 (37:00):
Okay.
Michelle, you raised your SeriesA two months after your seed
round.
Take us through that process.
SPEAKER_00 (37:09):
Yeah.
So we started the like raisingour seed round in November 2020.
All was finished in uh inJanuary.
By the way, we had to uh delaythe announcement because we're
trying to buy the domain.
And we didn't want to pay thepremium of uh being funded.
SPEAKER_03 (37:29):
So you had your round, you just didn't have the
domain.
Did you have a website on adifferent domain?
SPEAKER_00 (37:34):
Yes, we did.
SPEAKER_03 (37:35):
Okay, but you're trying to get the main domain
for the announcement.
How did you get it?
Did you have to just work outthe money or did you go
negotiate?
SPEAKER_00 (37:41):
No, but we it we we had it for like a a a good uh a
good price.
Okay.
Perfect.
You got it.
Secure lots of tractation herein front.
Um and the thing that happenedafter we actually announced our
series uh our seed, this is thefirst time we had put our slides
(38:02):
um live.
And it really created a lot andlot of traction on the open
source product.
Like people, because it wasreally solving a very, very,
very painful problem for thataudience.
And our numbers went likethrough the roof between like
(38:23):
January and May.
And that's also when we startedto build the engine to make sure
that contributors could be alsoinvolved in the project.
Before it was just us buildingbecause there was a lot of
foundational work that needed todo.
But we opened up the repo forexternal contribution.
I don't know, it was aroundMarch or beginning of April, and
(38:46):
it picked up really fast.
And I think at that point, whenyou see an industry that is
moving so fast, like data, uh atthe time it was not even AI, it
was just data, you see thatboom, suddenly we are present in
like 5,000.
Um I don't think it was a littlebit less.
Um, it was maybe a thousand uhdifferent companies after just
(39:08):
releasing the the repo for likea few months.
That creates a lot of uh ofattention.
And I think it's a veryinnovative way of like solving
the problem of how do you movedata around.
So that's uh that was uh I thinkthat I think they did a good
move, like going for like and wedid a good move on uh on on
(39:28):
raising the series A here, andit also allowed us to just
invest more into growing thecommunity.
SPEAKER_03 (39:34):
So are there any drawbacks to doing that?
Yes.
A lot of companies areevaluating timeline, and we
speak to many companies andadvise them around timelines,
and two months is very quick.
SPEAKER_00 (39:47):
Yeah.
SPEAKER_03 (39:47):
Now, what are the drawbacks or the pros to doing
so?
SPEAKER_00 (39:50):
Well, the one that is very simple is when you
release um an open sourceproject, you don't have you
don't get paid for it.
Yeah.
So the drawback is that uhsuddenly it puts you on uh on um
(40:13):
the expectations are high.
That's that I would say that'spretty much it.
But at the same time, you know,when we when we raise the series
A, and even when we raise theseed, we chatted with these
investors, and all the time wewere pick we were picking the
ones that had a very deepunderstanding of what it what it
(40:36):
means to build open source.
What it does it means to buildan open source company.
Because you don't do open sourcefor the sake of doing open
source, you do it because youhave a strategy.
And ours was very strongbottom-up awareness, building a
standard, and those can take alittle bit of time.
You know, you can look at youcan look at Elastic, you can
look at Ashi Corp, etc.
etc.
(40:56):
Like all of these, like youcreate a very strong base, yeah,
and then you figure out like allthe different basically your
real product market fit.
Um, and so I would say like notlike that's a risk of drawback.
We did not have it because wehad a a very uh knowledgeable um
uh investor on that front.
SPEAKER_03 (41:17):
Got it.
So it sounds like a learning foranyone thinking about this kind
of strategy or even just overallwith the alignment around
expertise with your investor.
SPEAKER_00 (41:27):
Exactly.
Okay.
The partner you're working with,well, yeah, they're gonna be
here for a very long time.
You better be very aligned withthem on like what you want to do
and also like their tolerancefor yes, things don't always go
right.
SPEAKER_03 (41:44):
And how do you evaluate that from the founder
seat?
Because naturally we evaluate itall the time from the other
side.
SPEAKER_00 (41:49):
Yeah.
Um, well, like always, when youin a way you you recruit
someone, yeah, back channels isthe best way.
So you talk to other companies,you you you search for the
company where it went well, theone that where it didn't go
well, and create a relationshipwith the with the people that
have been working there and andsee what they have to say.
SPEAKER_01 (42:09):
So excellent.
SPEAKER_00 (42:11):
And also you see, you know, you you also see like
during the during the thefundraising process, like is how
much are they um evolving yourthinking?
Uh, you know, when we raisedwith Axel.
SPEAKER_01 (42:27):
Yeah.
SPEAKER_00 (42:28):
Like I remember spending like two or three hours
with uh with Amit at the time,and he asked questions that in a
way helped us improve how wewere thinking about the the
future, the positioning ofAirbag, and what to do.
So there was already some verystrong value on like working
with uh with him or working withuh with Shetan at benchmark.
(42:50):
It's like they they help youthink.
And yes, they have theiropinion, I have my opinions, but
at the end of the day, like arethey allowing you to see places
what you don't know about?
SPEAKER_01 (43:04):
Right.
SPEAKER_00 (43:04):
And if so, I think that's a that can become a great
partnership.
SPEAKER_03 (43:08):
Great advice for anyone listening, thinking about
that investor founderrelationship.
Okay, take us back to a littlebit earlier.
If we were to circle back toyour product market fit, were
there any kind of biggestchallenges to that?
I think there were some big kindof moments around that.
SPEAKER_00 (43:30):
Um yes.
Um I would say like in 2022,that's when we we started to
work on the on the cloudproduct, which by the way, if
you're an open like for opensource founders, like going from
an open source product to anactual cloud product, it is
super hard.
(43:51):
Because hosting and managingsomething when what you've done
is like providing something thatyou don't need to really host
and manage, etc.
etc., this is very, very hard.
And in 2022, we released likethe let's call it like the
private beta of uh of AirbiteCloud.
There was a ton of problems, andwhich by the way is completely
(44:14):
normal, but we underestimatedhow complex it was to build a
platform.
And because we had this big planof like how we're gonna be
monetizing airbite, etc.
etc., I would say we we hired alittle bit ahead.
And that to me was uh was amistake because it also creates
a lot of noise internally.
(44:35):
It like disrupts the productteam, it disrupts engineering,
it creates like a lot of noisearound like building the best
product.
And that to me was uh I don'tknow, I would say was a bad
decision.
Uh we we had to course correct,but I would say it's like
(44:56):
especially when you're startingsomething new, just start small,
expand rather than go gobottom-up in terms of how you're
building your your your companyand your organization rather
rather than top-down.
There is a moment when you cando top-down when you have like a
lot more predictability, but atthe beginning, it's uh it's a
mistake.
SPEAKER_03 (45:14):
So bottom up.
SPEAKER_00 (45:15):
At least for us it was a mistake.
SPEAKER_03 (45:16):
Yeah, well, it sounds like in everything that
you did, you were always lookingat signals.
Like I know I've heard you saythat open source allowed you to
have signal density.
And through the communityaspect, you're talking about
using that as signals andfeedback too, and same with this
bottoms-up approach.
SPEAKER_00 (45:31):
Yeah, yeah.
I'm a I'm a very bottom-upperson on that front.
But at some point, yes, when Isee it's always the same thing,
like we all build we're buildingan engine.
So we need to figure out what isthe the MVP of that engine.
unknown (45:46):
Yeah.
SPEAKER_00 (45:46):
And to do that, you need to find people that are
extremely driven, that are okaywith uncertainty.
But the moment you start gettinglike an initial version that is
working, that's when you canstart like putting more uh like
more thought into what it shouldactually look like.
But first you need to validatesomething.
Definitely.
SPEAKER_03 (46:05):
Well, this has been fantastic, Michelle.
Really, really appreciate thetime.
A couple last questions.
You know, you have learnedimmensely throughout the
journey, but are there any booksthat you've particularly been
influenced by throughout yourcareer and life?
SPEAKER_00 (46:20):
Yeah, you see I don't know if you remember, but
I said giving a window into howthings are working to the
outside world.
Yeah, I did not invent that.
It's like when I was I think itwas in 2014, I was just um, or
2013, I was just starting to uhto manage my my first team, my
first team at the time.
(46:41):
And my CTO gave me this bookfrom um it's called Um High
Output Management.
I think it's now it's astandard.
Uh and it really, you know, whenyou go from being an IC to
starting to manage people, it'svery hard to find like the right
feedback clue for like, are youdoing a good job or not?
(47:02):
Like what does it mean thatyou're doing a good job?
And also how do you build teamas systems?
And I think that book was justtransformational for me because
I like good theory, yeah, andthat theory was very very
strong, and like how you create,how you how you build, how you
build the system, how youmonitor these systems, and um
(47:25):
and uh how you take pride of thework when you're not the one
always doing the work yourself.
SPEAKER_03 (47:31):
Yeah.
So great, fantastic.
Well, that will be in the shownotes for anyone curious to take
read and pay it for it a littlebit.
Where can people follow alongyour journey in airbytes?
SPEAKER_00 (47:40):
Uh well, the I would say the entry point is always
gonna be airbite.com.
SPEAKER_03 (47:45):
There we go.
SPEAKER_00 (47:46):
I'm on I'm on LinkedIn, I try to post as much
as I can.
SPEAKER_03 (47:49):
Yeah.
SPEAKER_00 (47:50):
Uh content marketing.
There we go.
You're very good at it.
And giving and giving a windowto uh to the to the people uh on
what we're doing.
And and yeah, and after that,like you can go on Slack, on our
GitHub repository, and just orjust try the product.
SPEAKER_03 (48:06):
There we go.
There's many ways.
Many ways.
SPEAKER_00 (48:08):
Point of entry is going to be the website.
Perfect.
SPEAKER_03 (48:11):
The website itself.
Well, Michelle, this has beenfabulous.
Really appreciate the time andyou sharing your journey and
insights.
SPEAKER_00 (48:15):
Yeah, thank you for having me.
It was a great conversation.
SPEAKER_03 (48:18):
Absolutely.
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
Ruthie's Table 4
For more than 30 years The River Cafe in London, has been the home-from-home of artists, architects, designers, actors, collectors, writers, activists, and politicians. Michael Caine, Glenn Close, JJ Abrams, Steve McQueen, Victoria and David Beckham, and Lily Allen, are just some of the people who love to call The River Cafe home. On River Cafe Table 4, Rogers sits down with her customers—who have become friends—to talk about food memories. Table 4 explores how food impacts every aspect of our lives. “Foods is politics, food is cultural, food is how you express love, food is about your heritage, it defines who you and who you want to be,” says Rogers. Each week, Rogers invites her guest to reminisce about family suppers and first dates, what they cook, how they eat when performing, the restaurants they choose, and what food they seek when they need comfort. And to punctuate each episode of Table 4, guests such as Ralph Fiennes, Emily Blunt, and Alfonso Cuarón, read their favourite recipe from one of the best-selling River Cafe cookbooks. Table 4 itself, is situated near The River Cafe’s open kitchen, close to the bright pink wood-fired oven and next to the glossy yellow pass, where Ruthie oversees the restaurant. You are invited to take a seat at this intimate table and join the conversation. For more information, recipes, and ingredients, go to https://shoptherivercafe.co.uk/ Web: https://rivercafe.co.uk/ Instagram: www.instagram.com/therivercafelondon/ Facebook: https://en-gb.facebook.com/therivercafelondon/ For more podcasts from iHeartRadio, visit the iheartradio app, apple podcasts, or wherever you listen to your favorite shows. Learn more about your ad-choices at https://www.iheartpodcastnetwork.com
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.