May 27, 2026 • 10 mins
I spoke with Kyle Kelly, the Founder and CEO of Zubu AI, which is an in-place data collection and preservation tool for litigation. We discussed the importance of in-place preservation, how Zubu is built for an emerging enterprise AI data layer, and what legal data preservation looks like within ChatGPT and Claude.
Listen
Watch
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:01):
Welcome to Reinventing Professionals,a podcast hosted by industry analyst
Ari Kaplan, which shares ideas,guidance, and perspectives from market
leaders shaping the next generationof legal and professional services.
This is Ari Kaplan, and I'm speakingtoday with Kyle Kelly, the founder
(00:24):
and CEO of Zubu AI, which isan in-place data collection and
preservation tool for litigation.
Hi, Kyle.
How are you?
Hey, Ari.
I'm doing well.
How are you doing?
I'm doing great.
I'm looking forward to this conversation.
So tell us about your backgroundand the genesis of Zubu.
I started my career doing criminal,computer, and cellphone forensics.
(00:49):
Really interesting work, providingsecond opinion on state's
evidence for defense attorneys.
A lot of high-profile crimes.
Got to see a lot of things, be in themidst of doing state's attorney general
work, working with them, getting deposed,being on stand to present evidence.
Super interesting work.
It also introduced me towhat e-discovery is today.
(01:10):
So I trans- positioned my forensics workinto e-discovery and worked at a small
cloud company you may have heard ofcalled Salesforce and a cryptocurrency
exchange called Coinbase you may haveheard of where I had the opportunity
to build out programs from scratchsupporting emerging technology systems
in both just general CRM and high tech,as well as looking at, the crypto market,
(01:33):
back then and what it has become today.
I had the opportunity to see a lotof the tools in the market throughout
that that experience, and I've beenable to translate that to identifying
the problems that are in the marketand the gaps that are, been there.
And after my tenure at Coinbase,I decided that I had my original
idea on how to think about howto solve some of those gaps.
One of that was actually on theright hand of the EDRM model, which
(01:56):
was how we're recycling data forlitigation once review comes around
and see how we are able to treat that.
, That ended up translating intomoving to the far left of the EDRM
model wondering once that idea wasa little too early for what the
market was ready for at that time.
And that became Zubu.
Zubu's actually inspired by the law.
(02:16):
For those who might remember, Zubulakeversus UBS Warburg, paramount case and the
requirement for you to have to preservedata on the anticipation of litigation.
That's where I started, and I've beenbuilding back to that original idea,
and we have been building in-placepres- preservation, data collection
to support those with litigation andit's been quite the adventure, Ari.
(02:37):
What was the moment when yourealized that the e-discovery
industry needed a tool like this?
If you think about legacy tools,they were built for a different era.
The assumptions baked into themno longer reflect how enterprise
data actually lives and moves.
The gap was not a feature gap necessarily.
It was more of an architectural one.
(02:58):
Not one that had builtfrom the foundation.
The opportunity was clear that thesetraditional platforms that, basically
build these big silos and copy andingest data for their own environment,
they create chain of custody risksand storage costs and data sprawl.
When we think about in place, wereally mean that the data never
leaves where it lives, right?
So Zubu preserves it at the source,inside of Google, Microsoft, Slack,
(03:20):
Confluence, and so at that pointwe realized that the people were
looking at this a little differently.
We continued on that track and saidhow else can we invigorate that
experience?" So we brought out thingslike BYOB, bring your own blob.
So that means that we believein complete data sovereignty.
Customers bring their own cloudstorage bucket, it stays in their own
(03:41):
infrastructure under their control.
Your data, your keys, your control.
For legal teams, that means, faster waysto be able to pre- preserve that data.
We lower that risk significantly andprovide a defensible preservation
record that without the overheadof massive ingestion pipeline.
And we think that's the way of the future,so that again, data lives in these silos.
Let them live and breathe , in thosesilos, and let them be interconnected
(04:04):
under a single pane of glass.
Why does in-place preservation matter?
In place means that you bring theownership back to the client, there's not
just, "Hey, okay, so I've sent it off tothis third party," whose third party has
a third party, who has a third party.
So all of a sudden you don'tactually really know who has
access to all of your data.
(04:24):
Now, over time it's gotten better, butin-place preservation means that you just
give the control back to the customer.
They don't have to worry aboutit being segmented, and this
is actually really relevant.
More than ever, it's so important tohave that ownership control internally.
That's why in-place preservation matters.
How is your platform built foran enterprise AI data layer?
(04:51):
The first generation ofcustodian data was email.
The second was the collaborationlayer, Slack, Teams, Google Drive.
We're now entering the third phase,the enterprise AI generation layer.
As you can imagine, employees are makingdecisions and drafting strategies that
are producing work products inside oftools like ChatGPT, Copilot, Claude, and
(05:13):
Gemini, and we're getting to the pointwhere this is becoming legally relevant.
We're needing to understand whatthis actually means for corporations.
What I've realized is thatwe're a little early on this.
Most legal teams have notnecessarily caught up to this
yet, but Zoome is building for itnow before it becomes a crisis.
We're understanding that the questionis not whether the AI platform data
will become discoverable, but morewhen, and we want to help teams be
(05:37):
prepared and ready for when it is
What does legal data preservation looklike within platforms like ChatGPT
Enterprise, Claude for Enterprise,and what should teams do to prepare?
So these AI platforms I will sayat this point all of them have
a compliance API of some sort,but they're not treated equally.
(05:57):
They have different functionalities,different ways of doing preservation.
The way I think about it isthese AI platforms are generating
conversations, logs, prompts andoutputs in, and in shared workspaces.
All of it, is likely to be really relevantfor litigation and investigation context.
So the Zuba integration , is going tobe designed to apply the same in-place
(06:19):
preservation model to AI data thatapplies to email and collaboration today.
So how do we look at that?
Teams should start by understandingthat what the AI tools risk are,
and, what their use access across theorganization have for export capabilities
and compliance responsibilities.
Governance needs to be aheadof the adoption necessarily.
(06:40):
That's where we're trying to be inthat governance layer for that in-place
preservation and in-place data collection.
So that legal team should have aseat at the table before the AI tool
rolls out, not after necessarily.
So we're hoping to help on theprocess side of things as well.
But how we're engaging with that datais those prompts, those conversations,
those exports at the point of creation,and be able to provide a way that a
(07:05):
layer in which that you're able topreserve that, understand that data
natively before it has to go out toa third party and be processed across
the board so that you have full controland visibility, and it also helps
your legal strategy going forward.
Your internal team knows what this datamight mean and that you're able to craft
better strategy when you're talkingabout how we're looking at this data,
(07:27):
what this data means for that particularmatter and how to move forward with more
intelligence about your AI data set.
What does e-discovery look likewhen it's built for where data's
going and not where data's been?
It looks like preservation thatdoes not require moving data.
Collection that does notrequire an army of consultants.
(07:52):
Holds that go out in minutes,necessarily not in hours or days.
It looks like a platform thattreats AI data as a first-class
data source, not an afterthought.
So you're not being anchoredto legacy assumptions that is
actually a competitive advantage.
Zuba is not built to the existingmodel, it's built to replace it.
(08:12):
Everyone's worried about replacing people,but the next generation of e-discovery,
will be redefined by those who build wheredata's going, not where it is, and that is
the only question that matters right now
How do you see e-discovery evolving?
Before Zubu, I was building somethingcalled Prediscovered, and that was looking
(08:33):
at the way that data was being treated.
So as we see, when review comes around,you've spent the money, you've had the
amount of review hours, which AI isnow automating in some capacity, at
least in the first pass review, andthen all that intelligence goes away.
Where I think where this will begoing eventually is that those
decisions that you paid expensivelawyers to make will eventually
(08:56):
turn into business intelligence.
That was the tagline, "Turning legalintelligence into business intelligence."
So I think in place is actually thefuture, and I've also, put my stake
in it, that's the tagline for Zubu.
The, the future ofe-discovery runs in place.
What I think the future's goingto be is that these silos will
be the ones that you've alreadytrusted from the very beginning.
They'll be living in Microsoft,in Slack, and Claude, and GPT.
(09:20):
And I think the discovery willbe that will be the layer.
Either tools will be as like the meteredlayer in between so that they communicate
between those different data sources,but I believe that those sort of legacy
silos will have to adapt to discoverybeing everything in place and having
high level visibility of What I thinkabout in terms of the file transparency.
(09:44):
What's the data lifecycle actually look like?
The moment that a Google Doc iscreated, now it was preserved in
Vault, and then it's been sent over tooutside counsel or third party vendor.
They've done their review.
All of this information will then belike a transmitted layer, and the more
that we cede the visibility, the morethat it'll actually enclose on itself,
and that'll actually be done in place.
(10:06):
This is Ari Kaplan speaking with KyleKelley, the founder and CEO of Zubu AI,
which is an in-place data collectionand preservation tool for litigation.
Kyle, thanks so very much.
Thank you so much, Ari.
Really appreciate the time.
(10:27):
Thank you for listening to theReinventing Professionals podcast.
Visit reinventingprofessionals.com orarikaplanadvisors.com to learn more.
Welcome to Reinventing Professionals,a podcast hosted by industry analyst
Ari Kaplan, which shares ideas,guidance, and perspectives from market
leaders shaping the next generationof legal and professional services.
This is Ari Kaplan, and I'm speakingtoday with Kyle Kelly, the founder
(00:24):
and CEO of Zubu AI, which isan in-place data collection and
preservation tool for litigation.
Hi, Kyle.
How are you?
Hey, Ari.
I'm doing well.
How are you doing?
I'm doing great.
I'm looking forward to this conversation.
So tell us about your backgroundand the genesis of Zubu.
I started my career doing criminal,computer, and cellphone forensics.
(00:49):
Really interesting work, providingsecond opinion on state's
evidence for defense attorneys.
A lot of high-profile crimes.
Got to see a lot of things, be in themidst of doing state's attorney general
work, working with them, getting deposed,being on stand to present evidence.
Super interesting work.
It also introduced me towhat e-discovery is today.
(01:10):
So I trans- positioned my forensics workinto e-discovery and worked at a small
cloud company you may have heard ofcalled Salesforce and a cryptocurrency
exchange called Coinbase you may haveheard of where I had the opportunity
to build out programs from scratchsupporting emerging technology systems
in both just general CRM and high tech,as well as looking at, the crypto market,
(01:33):
back then and what it has become today.
I had the opportunity to see a lotof the tools in the market throughout
that that experience, and I've beenable to translate that to identifying
the problems that are in the marketand the gaps that are, been there.
And after my tenure at Coinbase,I decided that I had my original
idea on how to think about howto solve some of those gaps.
One of that was actually on theright hand of the EDRM model, which
(01:56):
was how we're recycling data forlitigation once review comes around
and see how we are able to treat that.
, That ended up translating intomoving to the far left of the EDRM
model wondering once that idea wasa little too early for what the
market was ready for at that time.
And that became Zubu.
Zubu's actually inspired by the law.
(02:16):
For those who might remember, Zubulakeversus UBS Warburg, paramount case and the
requirement for you to have to preservedata on the anticipation of litigation.
That's where I started, and I've beenbuilding back to that original idea,
and we have been building in-placepres- preservation, data collection
to support those with litigation andit's been quite the adventure, Ari.
(02:37):
What was the moment when yourealized that the e-discovery
industry needed a tool like this?
If you think about legacy tools,they were built for a different era.
The assumptions baked into themno longer reflect how enterprise
data actually lives and moves.
The gap was not a feature gap necessarily.
It was more of an architectural one.
(02:58):
Not one that had builtfrom the foundation.
The opportunity was clear that thesetraditional platforms that, basically
build these big silos and copy andingest data for their own environment,
they create chain of custody risksand storage costs and data sprawl.
When we think about in place, wereally mean that the data never
leaves where it lives, right?
So Zubu preserves it at the source,inside of Google, Microsoft, Slack,
(03:20):
Confluence, and so at that pointwe realized that the people were
looking at this a little differently.
We continued on that track and saidhow else can we invigorate that
experience?" So we brought out thingslike BYOB, bring your own blob.
So that means that we believein complete data sovereignty.
Customers bring their own cloudstorage bucket, it stays in their own
(03:41):
infrastructure under their control.
Your data, your keys, your control.
For legal teams, that means, faster waysto be able to pre- preserve that data.
We lower that risk significantly andprovide a defensible preservation
record that without the overheadof massive ingestion pipeline.
And we think that's the way of the future,so that again, data lives in these silos.
Let them live and breathe , in thosesilos, and let them be interconnected
(04:04):
under a single pane of glass.
Why does in-place preservation matter?
In place means that you bring theownership back to the client, there's not
just, "Hey, okay, so I've sent it off tothis third party," whose third party has
a third party, who has a third party.
So all of a sudden you don'tactually really know who has
access to all of your data.
(04:24):
Now, over time it's gotten better, butin-place preservation means that you just
give the control back to the customer.
They don't have to worry aboutit being segmented, and this
is actually really relevant.
More than ever, it's so important tohave that ownership control internally.
That's why in-place preservation matters.
How is your platform built foran enterprise AI data layer?
(04:51):
The first generation ofcustodian data was email.
The second was the collaborationlayer, Slack, Teams, Google Drive.
We're now entering the third phase,the enterprise AI generation layer.
As you can imagine, employees are makingdecisions and drafting strategies that
are producing work products inside oftools like ChatGPT, Copilot, Claude, and
(05:13):
Gemini, and we're getting to the pointwhere this is becoming legally relevant.
We're needing to understand whatthis actually means for corporations.
What I've realized is thatwe're a little early on this.
Most legal teams have notnecessarily caught up to this
yet, but Zoome is building for itnow before it becomes a crisis.
We're understanding that the questionis not whether the AI platform data
will become discoverable, but morewhen, and we want to help teams be
(05:37):
prepared and ready for when it is
What does legal data preservation looklike within platforms like ChatGPT
Enterprise, Claude for Enterprise,and what should teams do to prepare?
So these AI platforms I will sayat this point all of them have
a compliance API of some sort,but they're not treated equally.
(05:57):
They have different functionalities,different ways of doing preservation.
The way I think about it isthese AI platforms are generating
conversations, logs, prompts andoutputs in, and in shared workspaces.
All of it, is likely to be really relevantfor litigation and investigation context.
So the Zuba integration , is going tobe designed to apply the same in-place
(06:19):
preservation model to AI data thatapplies to email and collaboration today.
So how do we look at that?
Teams should start by understandingthat what the AI tools risk are,
and, what their use access across theorganization have for export capabilities
and compliance responsibilities.
Governance needs to be aheadof the adoption necessarily.
(06:40):
That's where we're trying to be inthat governance layer for that in-place
preservation and in-place data collection.
So that legal team should have aseat at the table before the AI tool
rolls out, not after necessarily.
So we're hoping to help on theprocess side of things as well.
But how we're engaging with that datais those prompts, those conversations,
those exports at the point of creation,and be able to provide a way that a
(07:05):
layer in which that you're able topreserve that, understand that data
natively before it has to go out toa third party and be processed across
the board so that you have full controland visibility, and it also helps
your legal strategy going forward.
Your internal team knows what this datamight mean and that you're able to craft
better strategy when you're talkingabout how we're looking at this data,
(07:27):
what this data means for that particularmatter and how to move forward with more
intelligence about your AI data set.
What does e-discovery look likewhen it's built for where data's
going and not where data's been?
It looks like preservation thatdoes not require moving data.
Collection that does notrequire an army of consultants.
(07:52):
Holds that go out in minutes,necessarily not in hours or days.
It looks like a platform thattreats AI data as a first-class
data source, not an afterthought.
So you're not being anchoredto legacy assumptions that is
actually a competitive advantage.
Zuba is not built to the existingmodel, it's built to replace it.
(08:12):
Everyone's worried about replacing people,but the next generation of e-discovery,
will be redefined by those who build wheredata's going, not where it is, and that is
the only question that matters right now
How do you see e-discovery evolving?
Before Zubu, I was building somethingcalled Prediscovered, and that was looking
(08:33):
at the way that data was being treated.
So as we see, when review comes around,you've spent the money, you've had the
amount of review hours, which AI isnow automating in some capacity, at
least in the first pass review, andthen all that intelligence goes away.
Where I think where this will begoing eventually is that those
decisions that you paid expensivelawyers to make will eventually
(08:56):
turn into business intelligence.
That was the tagline, "Turning legalintelligence into business intelligence."
So I think in place is actually thefuture, and I've also, put my stake
in it, that's the tagline for Zubu.
The, the future ofe-discovery runs in place.
What I think the future's goingto be is that these silos will
be the ones that you've alreadytrusted from the very beginning.
They'll be living in Microsoft,in Slack, and Claude, and GPT.
(09:20):
And I think the discovery willbe that will be the layer.
Either tools will be as like the meteredlayer in between so that they communicate
between those different data sources,but I believe that those sort of legacy
silos will have to adapt to discoverybeing everything in place and having
high level visibility of What I thinkabout in terms of the file transparency.
(09:44):
What's the data lifecycle actually look like?
The moment that a Google Doc iscreated, now it was preserved in
Vault, and then it's been sent over tooutside counsel or third party vendor.
They've done their review.
All of this information will then belike a transmitted layer, and the more
that we cede the visibility, the morethat it'll actually enclose on itself,
and that'll actually be done in place.
(10:06):
This is Ari Kaplan speaking with KyleKelley, the founder and CEO of Zubu AI,
which is an in-place data collectionand preservation tool for litigation.
Kyle, thanks so very much.
Thank you so much, Ari.
Really appreciate the time.
(10:27):
Thank you for listening to theReinventing Professionals podcast.
Visit reinventingprofessionals.com orarikaplanadvisors.com to learn more.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by Audiochuck Media Company.
NFL Daily with Gregg Rosenthal
Gregg Rosenthal and a rotating crew of elite NFL Media co-hosts, including Patrick Claybon, Colleen Wolfe, Steve Wyche, Nick Shook and Jourdan Rodrigue of The Athletic get you caught up daily on all the NFL news and analysis you need to be smarter and funnier than your friends.