May 8, 2026 • 11 mins
I spoke with Katya Linossi and Gabriel Karawani, the Co-Founders of ClearPeople, the maker of Atlas, an intelligent knowledge platform for Microsoft 365 and the knowledge layer for precise, scalable AI. We discussed why consistency, context, and authority are the foundations of AI-ready knowledge, how a knowledge layer improves the accuracy of AI outputs, and ways that law firm leaders can ensure their AI tools use their firm's full knowledge base.
Listen
Watch
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
this is Ari Kaplan, and I'm speaking todaywith Katja Lenassi and Gabriel Karawani,
the co-founders of ClearPeople, themaker of Atlas, an intelligent knowledge
platform for Microsoft 365, and theknowledge layer for precise, scalable AI.
Katja, Gabriel, great to see you.
(00:20):
Thank you for having us on today.
Great to see you, Ari.
Looking forward to this conversation.
Really appreciate the time.
Katja, tell us about the genesis of
ClearPeople.
ClearPeople, started out as atechnical agency/consultancy.
We actually went on to becomelisted on the eConsultancy
100, so we grew quite big.
(00:41):
Then around about 2019, we decidedto pivot to create a product
called Atlas, the market-leadingintelligent knowledge platform.
We've been working with the legalsector since around about 2004.
In fact, I even did my MBA focusedon knowledge management in legal.
I also contributed a chapter tothe book called The Future of Legal
Knowledge Management (01:01):
Harnessing AI.
What inspire that courseof study at that time?
I was already working with a law firm.
They were wanting to do aknowledge management platform.
This is SharePoint days,but pre-called SharePoint.
It was called WSS.
So it was one of my projects I was workingon, I decided to combine that with my MBA.
(01:21):
I thought it was a great topic to do, sobeen in that space for a very long time.
Gabriel, when you and Katja launchedClearPeople, what assumption
about knowledge management didyou challenge, and how is AI
affecting your initial perspective?
Originally we didn't challenge manyassumptions in the original days.
(01:41):
But what we explored and discovered aswe were delivering projects, was that
we were resolving and solving the sameissues again and again for organizations.
So what we started challengingwas, why is this so special for one
organization versus another one?
Why can't we double down on someof these core challenges like
tagging content, for instance?
(02:02):
Why can't we double down on thoseand then build them into a product?
Fast-forward to 2019 we sat down andsay, "We must be able to deliver this
in a much better way, in a productizedway for organizations at scale."
Can you give us some background onwhat that initial process was like?
We were a straight down the middlesystem integrator, Microsoft
(02:24):
SharePoint consultancy back then.
We would come in, through the discoveryphase spend a lot of time with customers
whiteboarding, trying to refine taxonomieswith the customers and then at some
point get to something where you say,"Okay, this is now what you want, Mr.
Customer, Mrs. Customer.
Let's get it implemented." And soa lot of time was spent upfront on
doing things that effectively feltlike, we were repeating ourselves
(02:48):
across different customers, andwe also felt we could be much more
prescriptive with our experience.
We felt there was a lot we could bringto the table and start the customers from
a much higher bar from day one, insteadof reinventing the wheel every time.
One of those things that we identifiedearly on that we could really help
customers resolve is the whole captureprocess, the curation process where
(03:11):
typically it would involve a lot ofhuman effort, and we felt we could do
that a lot better through a product.
Katja, you've said that consistency,context, and authority are the
foundations of AI-ready knowledge.
What do each of those elements looklike practically inside of a law firm?
They're fundamental to precise AI results.
(03:33):
So let's start with consistency.
So this for us means knowledgeis structured in the same way,
especially across the firm.
Content is basically tagged againstcommon metadata and taxonomy.
And this is important so that AI isnot guessing, for example, whether two
documents are the same kind of thing.
You're telling it which isthe the right one to look at.
(03:56):
Now, inside a law firm, that might meanyou're consistently tagging for practice
areas, jurisdiction, document type,meta type, client sensitivity, status,
and without that, both search and AIretrieval becomes unreliable, so it's
really critical Then let's go to context.
Context is what you could sayfor AI, and even for search,
(04:20):
preserves meaning, purpose, audiencesensitivity, and relationships.
In a law firm, the right answer will,of course depend on your role, your
geography, your client, your matter,confidentiality, and even a use case.
A UK employment precedent youcould say is not interchangeable
with a US one, for example.
(04:41):
This is where Atlas is really strongbecause it delivers this contextual
and personalized knowledge based onyour role, your geography and this
could even be against how you'reusing this, whether you're looking
through this for an intranet, asearch experience, an extranet, or an
AI assistant, for example, as well.
And the third one is authority.
(05:02):
Authority is a signal that tells bothpeople and AI what it should trust.
And AI does not know what is best unlessyou tell it what it is, and it means
telling it what is approved content.
And even within the firm, it's thingslike clear ownership, life cycle
controls, review discipline and theability to distinguish what we call
(05:23):
gold standard from just your generalkind of information or draft material.
That's really critical, 'cause ifa lawyer or even an AI assistant is
presented with five similar documentsand none are clearly marked as approved
or current version, or even the goldstandard, you have a risk problem.
So this is something that we talk alot about because this is what we've
(05:44):
addressed in what we're doing in Atlasand how we do this through things like
authoritative knowledge collections,life cycle governance, and even the
ability to define what knowledge issafe for AI use, and I think this
is where it becomes really critical.
People can figure out their waythrough things, but AI doesn't
know how to do that, so it's reallycritical that you need to tell AI
(06:06):
what is authoritative, Give it thatcontent and then that consistency.
Gabriel, how does a knowledge layerimprove the accuracy of AI outputs as
well as consistency, trustworthiness,and of course defensibility?
Katja's the three pointsthere are fundamental to that.
It's interesting you use the termknowledge layer, because I don't
(06:28):
think many people were sayingknowledge layer a few years ago.
Now we actually get prospective customerscome to us and say, "Hey, we need a
knowledge layer." As a market, we don'treally all agree necessarily what that is.
But for us it's really aboutcontrolling, what is accessed through
the knowledge layer, what AI cansee through the knowledge layer.
(06:48):
And if you think about the evolutionof MCPs, so Model Context Protocol,
when you're exposing your knowledgeto external tools or to users
via external tools, it's not justnecessarily just about permissions
or about the freshness maybe of thecontent or the accuracy of the content.
It's also about that contextin which it is delivered.
(07:09):
So controlling what AI sees andhow it uses it is a fundamental
part of the knowledge layer.
The knowledge layer improvesoverall accuracy because it gives
the, cleaner, better structuredanswers than you otherwise would.
When we talk about a knowledge layer,we're not just talking about knowledge
within a set of clearly defined knowledgerepositories in one knowledge system.
(07:35):
We actually talk about a layerthat brings knowledge together
from potentially multiple systems.
It could be from iManage, it couldbe from SharePoint, it could be from
Salesforce, it could be from Adorant.
We can bring these things togetherinto one knowledge layer, and that
is what many are asking for becausesolutions today are very siloed.
(07:57):
You've got a fantastic solution frompoint provider X and another fantastic
solution from provider Y, but it'svery rare and very difficult to bring
these together in one knowledge layersolution where you can ask your question
once and get an answer based on yourauthoritative data from multiple systems.
(08:19):
Katja, how can law firm leaders ensurethat their AI tools use the full knowledge
base rather than siloed inconsistentpockets of knowledge at the firm?
It's about building a trusted knowledgebetween your content and your AI tools.
Today a lot of firms are optimizingtheir knowledge for a single AI tool, and
(08:40):
the problem is that this doesn't scale.
If AI expands across, the firm,you're likely to have over time the
same structural issues that reappearor you have inconsistent metadata,
you could have duplicate content,maybe even poor context retrieval.
So it's really important that firms thinkabout shifting from tool centric thinking,
(09:02):
to a foundational centric thinking.
That means defining what knowledgeis authoritative applying consistent
structure and metadata, across systems.
So you're creating this curatedpermission aware knowledge
collections that AI can safely use.
Firms need to think about designfor portability because they're
(09:22):
not gonna only have one AI tool.
So they need to have the knowledgelayer, so that all of the tools
they're using operate fromthat same governed foundation.
A really important point Katja'smaking there is that portability.
I've invested in AI tool X, butmaybe I've also got, say, Copilot.
Most law firms that would belistening to this, they've got
(09:43):
investments into Microsoft 365.
They'll be using Copilot.
But what if you could use Copilotas your entry point into that
knowledge layer through MCP?
This is the presentation of Atlas Fuse.
I'm sitting in a tool where actuallymy answers aren't that accurate until I
connect the tool, Copilot in this case,to the knowledge layer, and suddenly
(10:04):
I am getting highly accurate answers.
This is super interesting fororganizations to consider.
One entry point, but a knowledge layeracross multiple sources and multiple tools
that you could potentially use dependingon the use case or scenario, and I think
that's how we're gonna look at AI agents.
That's why you need this foundation,because AI agents might be doing
(10:25):
multiple things or differentthings, but they need to work from
consistency, and that goes back toconsistency, context, and authority.
Katja, for a firm that's earlyin its AI journey, what mindset
shift do leaders need to makeabout knowledge before investing
further in artificial intelligence?
(10:45):
It goes back to thinking aboutyour knowledge foundation.
We've seen this alsoacross other customers.
Instead of focusing on AI, theyfocus specifically on, how do we
get our knowledge foundation right?
How do we get our governance right?
Even cleaning up contentmade a huge difference.
When you get that knowledge foundationright, everything else came into play.
(11:08):
So a customer was on a webinarrecently and said how once they
had everything in place, everythingjust worked quickly and the search
better, the Copilot experience better.
Go back to that knowledgefoundation, those basics.
Get that right, and everythingelse scales from there.
You gotta get the foundations right.
But think about it from astrategic point of view.
Don't neglect the foundations.
(11:29):
Get that right.
Look at the overall firm objectivesand then everything else follows.
This is Ari Kaplan speaking with KatjaLinnassey and Gabriel Karawani, the
co-founders of ClearPeople, the makerof Atlas, an intelligent knowledge
platform for Microsoft 365 and theknowledge layer for precise, scalable AI.
(11:50):
Katja, Gabriel, thank you so very much.
Thank you so much for having us.
this is Ari Kaplan, and I'm speaking todaywith Katja Lenassi and Gabriel Karawani,
the co-founders of ClearPeople, themaker of Atlas, an intelligent knowledge
platform for Microsoft 365, and theknowledge layer for precise, scalable AI.
Katja, Gabriel, great to see you.
(00:20):
Thank you for having us on today.
Great to see you, Ari.
Looking forward to this conversation.
Really appreciate the time.
Katja, tell us about the genesis of
ClearPeople.
ClearPeople, started out as atechnical agency/consultancy.
We actually went on to becomelisted on the eConsultancy
100, so we grew quite big.
(00:41):
Then around about 2019, we decidedto pivot to create a product
called Atlas, the market-leadingintelligent knowledge platform.
We've been working with the legalsector since around about 2004.
In fact, I even did my MBA focusedon knowledge management in legal.
I also contributed a chapter tothe book called The Future of Legal
Knowledge Management (01:01):
Harnessing AI.
What inspire that courseof study at that time?
I was already working with a law firm.
They were wanting to do aknowledge management platform.
This is SharePoint days,but pre-called SharePoint.
It was called WSS.
So it was one of my projects I was workingon, I decided to combine that with my MBA.
(01:21):
I thought it was a great topic to do, sobeen in that space for a very long time.
Gabriel, when you and Katja launchedClearPeople, what assumption
about knowledge management didyou challenge, and how is AI
affecting your initial perspective?
Originally we didn't challenge manyassumptions in the original days.
(01:41):
But what we explored and discovered aswe were delivering projects, was that
we were resolving and solving the sameissues again and again for organizations.
So what we started challengingwas, why is this so special for one
organization versus another one?
Why can't we double down on someof these core challenges like
tagging content, for instance?
(02:02):
Why can't we double down on thoseand then build them into a product?
Fast-forward to 2019 we sat down andsay, "We must be able to deliver this
in a much better way, in a productizedway for organizations at scale."
Can you give us some background onwhat that initial process was like?
We were a straight down the middlesystem integrator, Microsoft
(02:24):
SharePoint consultancy back then.
We would come in, through the discoveryphase spend a lot of time with customers
whiteboarding, trying to refine taxonomieswith the customers and then at some
point get to something where you say,"Okay, this is now what you want, Mr.
Customer, Mrs. Customer.
Let's get it implemented." And soa lot of time was spent upfront on
doing things that effectively feltlike, we were repeating ourselves
(02:48):
across different customers, andwe also felt we could be much more
prescriptive with our experience.
We felt there was a lot we could bringto the table and start the customers from
a much higher bar from day one, insteadof reinventing the wheel every time.
One of those things that we identifiedearly on that we could really help
customers resolve is the whole captureprocess, the curation process where
(03:11):
typically it would involve a lot ofhuman effort, and we felt we could do
that a lot better through a product.
Katja, you've said that consistency,context, and authority are the
foundations of AI-ready knowledge.
What do each of those elements looklike practically inside of a law firm?
They're fundamental to precise AI results.
(03:33):
So let's start with consistency.
So this for us means knowledgeis structured in the same way,
especially across the firm.
Content is basically tagged againstcommon metadata and taxonomy.
And this is important so that AI isnot guessing, for example, whether two
documents are the same kind of thing.
You're telling it which isthe the right one to look at.
(03:56):
Now, inside a law firm, that might meanyou're consistently tagging for practice
areas, jurisdiction, document type,meta type, client sensitivity, status,
and without that, both search and AIretrieval becomes unreliable, so it's
really critical Then let's go to context.
Context is what you could sayfor AI, and even for search,
(04:20):
preserves meaning, purpose, audiencesensitivity, and relationships.
In a law firm, the right answer will,of course depend on your role, your
geography, your client, your matter,confidentiality, and even a use case.
A UK employment precedent youcould say is not interchangeable
with a US one, for example.
(04:41):
This is where Atlas is really strongbecause it delivers this contextual
and personalized knowledge based onyour role, your geography and this
could even be against how you'reusing this, whether you're looking
through this for an intranet, asearch experience, an extranet, or an
AI assistant, for example, as well.
And the third one is authority.
(05:02):
Authority is a signal that tells bothpeople and AI what it should trust.
And AI does not know what is best unlessyou tell it what it is, and it means
telling it what is approved content.
And even within the firm, it's thingslike clear ownership, life cycle
controls, review discipline and theability to distinguish what we call
(05:23):
gold standard from just your generalkind of information or draft material.
That's really critical, 'cause ifa lawyer or even an AI assistant is
presented with five similar documentsand none are clearly marked as approved
or current version, or even the goldstandard, you have a risk problem.
So this is something that we talk alot about because this is what we've
(05:44):
addressed in what we're doing in Atlasand how we do this through things like
authoritative knowledge collections,life cycle governance, and even the
ability to define what knowledge issafe for AI use, and I think this
is where it becomes really critical.
People can figure out their waythrough things, but AI doesn't
know how to do that, so it's reallycritical that you need to tell AI
(06:06):
what is authoritative, Give it thatcontent and then that consistency.
Gabriel, how does a knowledge layerimprove the accuracy of AI outputs as
well as consistency, trustworthiness,and of course defensibility?
Katja's the three pointsthere are fundamental to that.
It's interesting you use the termknowledge layer, because I don't
(06:28):
think many people were sayingknowledge layer a few years ago.
Now we actually get prospective customerscome to us and say, "Hey, we need a
knowledge layer." As a market, we don'treally all agree necessarily what that is.
But for us it's really aboutcontrolling, what is accessed through
the knowledge layer, what AI cansee through the knowledge layer.
(06:48):
And if you think about the evolutionof MCPs, so Model Context Protocol,
when you're exposing your knowledgeto external tools or to users
via external tools, it's not justnecessarily just about permissions
or about the freshness maybe of thecontent or the accuracy of the content.
It's also about that contextin which it is delivered.
(07:09):
So controlling what AI sees andhow it uses it is a fundamental
part of the knowledge layer.
The knowledge layer improvesoverall accuracy because it gives
the, cleaner, better structuredanswers than you otherwise would.
When we talk about a knowledge layer,we're not just talking about knowledge
within a set of clearly defined knowledgerepositories in one knowledge system.
(07:35):
We actually talk about a layerthat brings knowledge together
from potentially multiple systems.
It could be from iManage, it couldbe from SharePoint, it could be from
Salesforce, it could be from Adorant.
We can bring these things togetherinto one knowledge layer, and that
is what many are asking for becausesolutions today are very siloed.
(07:57):
You've got a fantastic solution frompoint provider X and another fantastic
solution from provider Y, but it'svery rare and very difficult to bring
these together in one knowledge layersolution where you can ask your question
once and get an answer based on yourauthoritative data from multiple systems.
(08:19):
Katja, how can law firm leaders ensurethat their AI tools use the full knowledge
base rather than siloed inconsistentpockets of knowledge at the firm?
It's about building a trusted knowledgebetween your content and your AI tools.
Today a lot of firms are optimizingtheir knowledge for a single AI tool, and
(08:40):
the problem is that this doesn't scale.
If AI expands across, the firm,you're likely to have over time the
same structural issues that reappearor you have inconsistent metadata,
you could have duplicate content,maybe even poor context retrieval.
So it's really important that firms thinkabout shifting from tool centric thinking,
(09:02):
to a foundational centric thinking.
That means defining what knowledgeis authoritative applying consistent
structure and metadata, across systems.
So you're creating this curatedpermission aware knowledge
collections that AI can safely use.
Firms need to think about designfor portability because they're
(09:22):
not gonna only have one AI tool.
So they need to have the knowledgelayer, so that all of the tools
they're using operate fromthat same governed foundation.
A really important point Katja'smaking there is that portability.
I've invested in AI tool X, butmaybe I've also got, say, Copilot.
Most law firms that would belistening to this, they've got
(09:43):
investments into Microsoft 365.
They'll be using Copilot.
But what if you could use Copilotas your entry point into that
knowledge layer through MCP?
This is the presentation of Atlas Fuse.
I'm sitting in a tool where actuallymy answers aren't that accurate until I
connect the tool, Copilot in this case,to the knowledge layer, and suddenly
(10:04):
I am getting highly accurate answers.
This is super interesting fororganizations to consider.
One entry point, but a knowledge layeracross multiple sources and multiple tools
that you could potentially use dependingon the use case or scenario, and I think
that's how we're gonna look at AI agents.
That's why you need this foundation,because AI agents might be doing
(10:25):
multiple things or differentthings, but they need to work from
consistency, and that goes back toconsistency, context, and authority.
Katja, for a firm that's earlyin its AI journey, what mindset
shift do leaders need to makeabout knowledge before investing
further in artificial intelligence?
(10:45):
It goes back to thinking aboutyour knowledge foundation.
We've seen this alsoacross other customers.
Instead of focusing on AI, theyfocus specifically on, how do we
get our knowledge foundation right?
How do we get our governance right?
Even cleaning up contentmade a huge difference.
When you get that knowledge foundationright, everything else came into play.
(11:08):
So a customer was on a webinarrecently and said how once they
had everything in place, everythingjust worked quickly and the search
better, the Copilot experience better.
Go back to that knowledgefoundation, those basics.
Get that right, and everythingelse scales from there.
You gotta get the foundations right.
But think about it from astrategic point of view.
Don't neglect the foundations.
(11:29):
Get that right.
Look at the overall firm objectivesand then everything else follows.
This is Ari Kaplan speaking with KatjaLinnassey and Gabriel Karawani, the
co-founders of ClearPeople, the makerof Atlas, an intelligent knowledge
platform for Microsoft 365 and theknowledge layer for precise, scalable AI.
(11:50):
Katja, Gabriel, thank you so very much.
Thank you so much for having us.
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.
Betrayal Weekly
Betrayal Weekly is back for a new season. Every Thursday, Betrayal Weekly shares first-hand accounts of broken trust, shocking deceptions, and the trail of destruction they leave behind. Hosted by Andrea Gunning, this weekly ongoing series digs into real-life stories of betrayal and the aftermath. From stories of double lives to dark discoveries, these are cautionary tales and accounts of resilience against all odds. From the producers of the critically acclaimed Betrayal series, Betrayal Weekly drops new episodes every Thursday. If you would like to share your story, you can reach out to the Betrayal Team by emailing them at betrayalpod@gmail.com and follow us on Instagram at @betrayalpod and @glasspodcasts. Please join our Substack for additional exclusive content, curated book recommendations, and community discussions. Sign up FREE by clicking this link Beyond Betrayal Substack. Join our community dedicated to truth, resilience, and healing. Your voice matters! Be a part of our Betrayal journey on Substack.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com