Latest / Justice Team Podcast / Harnessing AI for Legal Excellence with Muneeb Khadeer of Klip

Transcript

Harnessing AI for Legal Excellence with Muneeb Khadeer of Klip · Justice Team Podcast

0:02
Bob: Welcome to this episode of the Justice Team podcast
0:05
on the Justice Team Network.
0:06
And today we're talking about technology and artificial
0:08
intelligence and specifically ways to better manage your documents.
0:12
And as lawyers, we are essentially large language models because we do a lot of.
0:16
Large languages.
0:17
And we're very honored to have Muneeb Qadir on today, who is the CEO and
0:21
founder of a company called Klip, K L I P, which we are actively using at our firm.
0:27
So welcome on.
0:28
Muneeb: Thanks, Bob.
0:29
Thanks for having me.
0:30
Bob: Yeah.
0:30
So some of the things we'll jump into, just so you can understand what
0:32
we're going to hear on this episode.
0:34
is we're going to get into some practical use cases, where artificial
0:37
intelligence is going, how you pretty much will be able, as a lawyer,
0:42
need to know how to query, just a, essentially a chatbot, um, and how to
0:47
better use your data in your company.
0:50
Muneeb: Alright,
0:50
Bob: so why don't you give us a quick, you know, rundown of kind of
0:52
your education and how you got here.
0:54
Muneeb: Yeah, absolutely.
0:55
So, I'm a mechanical engineer by trade and I've been studying computer science
0:59
and I've gravitated over to that.
1:01
over a decade ago, and I've been working in technology and in the tech space
1:05
for over a decade, you know, working on products that use the latest technology.
1:10
Some use AI, some use heuristics, but I'm very excited about AI and have had a
1:14
lot of experience with it, so I decided to take a chance and start this company
1:20
last year, and it's been amazing so far.
1:22
Bob: Yeah, and my understanding is, You guys, your company's out
1:26
of beta now, and in short order already got SOC 2 compliance.
1:30
That's right.
1:30
Can you explain to our listeners what that is?
1:32
Muneeb: Yeah, so SOC 2 compliance is a standard, uh, that, you know, has been
1:36
established, which allows and tells the world that, hey, this company is
1:41
securing the data, it's keeping your data private, um, and it's something that,
1:45
you know, you can trust because they go through auditing, they make sure that
1:48
your data is encrypted, it's, it's secure so that, you know, you can be confident.
1:54
Bob: So if somebody hears those use cases, well like, hey, I use chat GBT.
1:59
I put my clients information out there and now their information is public record.
2:04
And now I, you know, I I, I leaked some confidential information.
2:07
Like, what do you have to say to those listeners that have that fear?
2:09
Muneeb: Yeah, so that's a very valid fear, right?
2:12
Because ChatGPT is an amazing, uh, application.
2:15
It kind of has started this whole trend and has got everything going.
2:19
But it's built for all the consumers in the world, right?
2:22
So, there's tons of data that's going in there.
2:24
tons of data being taken out.
2:25
So, uh, those models are being trained on a daily basis with that data.
2:30
So that data, if you read the agreements or read anything, is very
2:34
much going in and being reused, right?
2:36
I know they're working on trying to make it more secure and more private,
2:40
but it's gonna be very challenging just because they're Uh, bringing in all
2:43
the data from everybody in the world.
2:45
And
2:45
Bob: that's what we tell, you know, a lot of lawyers and law firms.
2:48
Don't put any records into an open source like ChatGBT.
2:52
Don't put any client's medical records or any confidential documents because it
2:56
could, it could get out to the public.
2:57
That's right.
2:58
And what's the difference between what you're doing at CLIP and
3:00
then what is that, that open ChatGBT which people know from?
3:04
Muneeb: Yes, so, Uh, what we're doing at Clip is we have our own architecture,
3:09
and we're using our own cloud services.
3:11
So we're using both Amazon and Azure.
3:13
So Microsoft also has a, um, you know, Azure API that we can use, uh, in order
3:19
to be able to leverage some of the GPD services, but in a private way.
3:23
Uh, at the same time, we're also using the large language models that are available
3:27
now on Amazon, Amazon Bedrock, right?
3:29
AWS Bedrock.
3:30
Uh, so we're actually using four different models, Uh, within our platform so that
3:35
we can make sure that we are able to, uh, give you the best results, right,
3:39
for the type of question you're asking.
3:41
Bob: So, you know, a lot of companies, technology companies, are, are built
3:45
on certain systems and infrastructure, where as opposed to CLIP is
3:49
built on artificial intelligence.
3:51
That's right.
3:51
So walk, walk everybody through the difference, because I think it's
3:54
going to be important how things play out in the future as to what
3:56
system your warehouse was built on.
3:59
Muneeb: Yeah, so one of the things early on that I was very adamant
4:03
about with this technology space.
4:04
A lot of people are building models, right?
4:06
So a lot of companies were coming up trying to build models.
4:09
Uh, you know, Stanford has a great website which kind of ranks all
4:12
the models that are available.
4:14
Um, so I knew that these models are going to continuously evolve.
4:17
So I didn't want to be tied down to one model or one API
4:19
for that one model, right?
4:21
So the architecture that we built Particularly use
4:23
is actually model agnostic.
4:25
So that's what allows us to use multiple models.
4:28
And as those models evolve and change, we can change those out right to be able
4:32
to provide the best value to the user based on the questions they're asking.
4:36
Bob: But how does that compare to somebody that has?
4:40
So essentially a
4:41
Muneeb: CRM
4:41
Bob: and it's not built with those models or any models.
4:45
How's that?
4:45
How's clip different from that?
4:46
Muneeb: Yeah.
4:47
So all of those softwares were built on traditional architecture, right?
4:50
So they're using, you know, maybe they're using SQL based databases
4:54
or they're using, um, you know, document databases, right?
4:57
So with the advent of A.
4:59
I.
4:59
There's new databases like there's vector databases now that we use, for
5:02
example, right, along with the large language models, which are putting all
5:07
of these words, all of these, uh, all this information, a vector cloud, right?
5:11
So and then being able to query that information so they
5:14
can also start doing that.
5:15
But they would have to essentially to get the best results, re
5:18
architect their software.
5:19
Bob: Yeah.
5:20
And, you know, for people that have been listening to the show
5:22
or, you know, see somebody speaking engagements I've been saying for.
5:25
almost two years now that lawyers are just going to need to know how to query
5:29
a bot or ask a question like they do on Google in order to get the right
5:33
answer with artificial intelligence.
5:35
And to me, that's what clip is.
5:36
And what it does is it is your it's relying on your internal documents,
5:40
whatever documents you want to rely on.
5:43
One thing that I was excited when I was researching the product is
5:46
they're a partner with Microsoft.
5:47
So you can Clip can plug into your SharePoint, your Microsoft documents,
5:52
even your Microsoft Outlook, and your other systems, so you can
5:55
ask it any question you want.
5:56
Muneeb: That's right.
5:57
Right?
5:57
That's right.
5:58
And we wanted to keep it very structured, right?
6:00
Because, uh, to your matter, or to the things that you're working on.
6:04
So even your emails, if they're related to, for example, your matter.
6:07
Uh, it would be able to understand and have all that information and be
6:10
digested so that when you're asking questions and you need information
6:14
on the go, it's able to answer it with just that context, right?
6:17
Not everything else.
6:18
Bob: How far do you think we are away from it being so intelligent that it
6:21
can answer like my emails and I just have to review it and click a button?
6:24
Muneeb: I think that's, that's possible today.
6:26
Wow.
6:27
Really?
6:27
Yeah, that's possible today.
6:28
Now the challenge is that, you know, I love to keep the human in the loop, right?
6:32
In this particular situation where.
6:34
It can come up with a draft, right, and show you the draft,
6:37
and then you can edit the draft.
6:38
So that also gives us feedback at the same time.
6:40
Well, let me, let
6:41
Bob: me ask you, this is interesting to me.
6:42
Imagine a world where a clip is tied into all of my documents.
6:45
It knows everything about that case, hypothetically, because we've
6:48
queried it so many times, that if somebody emails me about that
6:52
case, it can have a proposed clip.
6:55
Maybe even with documents attached to it, answer to that question, and I just have
6:59
to look at, review it, and press send.
7:00
Muneeb: That's right.
7:01
Wow.
7:02
Yeah, it's phenomenal what, and this is the whole power of AI, right?
7:06
It's not heuristics.
7:07
Like, heuristics meaning, you know, a solution where you're able to, Define
7:13
every use case right so that it gives that answer This is more a brain so
7:17
you can think of it even the way what we call is a neural network, right?
7:20
So it's almost just like you're how your brain processes information
7:24
When you get in information you hit certain neurons fire in your brain and
7:28
then you retrieve the information for your brain and then you say It right.
7:31
It's similarly designed right and that's the whole beauty of AI it has
7:35
that's why sometimes, you know When they really push it, you know Uh,
7:38
it can hallucinate and do different
7:40
Bob: things.
7:40
And that's why, like, you know, at my firm and other firms is, we
7:44
always wanted to give the correct answer to our firm, associates,
7:48
staff, on every particular issue.
7:50
And we're asked the same question a lot of times through training and these things.
7:53
You know, and sometimes you get bad information from outside sources.
7:57
My excitement with CLIP is it's your internal data set So it's taking all of
8:00
your internal work product and thought processes and getting an answer that way.
8:05
Muneeb: That's right
8:05
Bob: Right
8:06
Muneeb: and and one of the main things with me is
8:08
repeatability and accuracy, right?
8:09
So What I like to say is accurate and complete is like
8:14
the ultimate utopia, right?
8:16
and then there's accurate but incomplete which means as You know,
8:19
LLMs are getting better and better.
8:21
They'll be, they can be accurate, but they're not going
8:23
through all the data, right?
8:24
Because you have volumes and volumes of data, so that's
8:26
where the fine tuning comes in.
8:28
Then there's inaccurate and incomplete, which is like the worst situation.
8:32
Yeah, and that's
8:33
Bob: what we see.
8:34
And the other thing is, you know, with CLIP, it's also, I
8:36
would say, trust but verify.
8:37
You've got to have that hyperlink availability to go back to see
8:40
where did you get that answer from and make sure it's correct.
8:43
Muneeb: That's exactly right, and that's why one of the main
8:45
features we have is sources, right?
8:47
So that's your document, but it tells you the page number, tells you
8:51
where it got the information and be able, is going to be able to, uh,
8:54
you will be able to verify that.
8:56
Bob: Yeah.
8:56
So do you think that this means that law firms don't need staff,
9:00
it's going to replace their jobs?
9:02
Muneeb: So I think that, you know, it's, it's, it's about humans that are going to
9:06
be, humans with AI will replace people.
9:09
Will replace humans without AI that's kind of my vision here, right?
9:13
So I think it's not that they won't need staffs, but I feel that they'll
9:16
empower their staff to do more
9:17
Bob: Yeah, and that's the thing is I you know, we've talked about this ad
9:20
nauseum But it allows staff to do less menial tasks They have to repeat over
9:25
and over and actually use more brainpower to do Yeah, more things right so like
9:30
for me as a lawyer rather than sifting through all these things and doing the
9:33
same mundane Mundane thing all the time.
9:36
I have the AI assist to be in my client's house That's right to be
9:39
in the courtroom right to do more because I have this powerful tool
9:43
Muneeb: Yep, and it'll create a I would say it's gonna create a much
9:46
better experience for everyone Right being able to ask those questions
9:50
quickly get accurate answer and also not waste time right to the main thing
9:55
is wasting time and Some of these settlements people would like to settle
9:58
faster, right and get through quicker just because of the situation So I
10:03
think that's gonna be the main thing.
10:04
So
10:04
Bob: here let's do let's go over some use cases of clip, right?
10:07
So my firm gets sent a lot of cases from other firms that are already
10:11
in litigation or about to go to litigation or almost At trial to say
10:15
hey, can you guys try this case for us?
10:17
Can you litigate this case for us and a lot of our time is spent
10:20
Through high level brain power of our partners to sift through all these
10:23
files, to see if there's any reasons why we should not take the case.
10:27
or some landmines or, or et cetera.
10:30
How can CLIP help us?
10:31
Muneeb: So you can take all those documents that you have, create a matter
10:34
within CLIP, upload all those documents, and then based on your knowledge and
10:38
experience, start asking questions that you can quickly assess and determine
10:41
if those landmines are available in that, those documents, then you'd be
10:45
like, Hey, I don't want to take this.
10:46
And
10:46
Bob: that, I mean, for me, that's a huge time saver to say the quick, no, right.
10:50
Rather than, Four hours of attorney time to figure out that's a no.
10:54
Yes.
10:54
But could I come up with a solution where I know 20 essentially questions
10:58
that are pretty critical in most of the questions or cases that we review.
11:01
Yeah.
11:01
To see if I can get to standardize.
11:03
I already got this new case clip.
11:05
Do your thing and answer these 20 cases for 20 questions for
11:07
Muneeb: absolutely right and so that's part of the design of the user
11:11
interface, right so we can already pre coined that we can have that available.
11:15
So the moment you do that, you just go to a page which has all
11:18
that information already populated.
11:20
So you're not able to you don't have to actually ask those questions.
11:22
So
11:22
Bob: it could be like.
11:23
What's the policy limit, is there any prior medical treatment, what's the first
11:26
medical treatment, show me, show me the MRI findings, was there surgery, what was
11:31
the surgery date, what are the settlement offers, what are the demands, all that.
11:35
Absolutely.
11:36
It can do it.
11:36
Muneeb: Yeah, it can do that right now.
11:37
Hyperlinked.
11:38
Hyperlinked, where it already do it while it processes the documents, so
11:41
that when you actually just go to the details page, you'll be able to see that.
11:44
Bob: Yeah.
11:44
My brother always says he usually just to start with the defense medical
11:47
exam report and work backwards.
11:49
He's like, I could get a lot just by reviewing that medical report.
11:51
Um, but yeah, but even if it quickly finds that report, hyperlinks it, it's
11:55
a huge time save to be able to do that.
11:57
Um, so another use case of it, right?
11:59
So we're showing you before we came in here, I have a trial next week and
12:02
there's dozens of depositions, a liability dispute with a catastrophic injury.
12:06
And I think there's over a terabyte of information that we dumped into CLIP.
12:09
Muneeb: That's right.
12:09
Bob: So what are some ways that we can query your system and what are some things
12:13
that can do You know, quick and dirty or long term to help us in a case like that.
12:17
Muneeb: Yeah, so in that particular situation, right, that the
12:20
documents you put in, there's like close to a thousand documents.
12:23
They could vary, right?
12:24
So one of the things is it can help you understand the different types
12:28
of documents that you have, right?
12:30
And what it is that the information that you need specifically, by
12:33
being able to do that by itself, by going through the documents.
12:37
And right now we're supporting certain formats, but we can start doing more
12:41
Uh, vision is to do like, you know, videos, audio files, everything, right?
12:44
So that you have all that context.
12:47
So when you're actually asking questions, you're able to get the
12:49
answers at the same time, uh, be able to see the sources of it.
12:53
So whether you actually ask it, or we have an interface for you, like,
12:57
almost like a dashboard, right?
12:58
Which are some of the things that may help you quickly assess that case.
13:01
Bob: Yeah, and you know, the way that it works is, it's all OCR'd,
13:05
so that it's extracting all of the words, from PDFs and then your large
13:10
language model is Picking that up at hyper speed to be able to answer the
13:15
questions over a thousand documents.
13:17
Muneeb: That's right And you know, there are
13:18
Bob: thousands and thousands of pages
13:19
Muneeb: thousands of pages I actually I would say probably
13:22
millions of pages, right?
13:23
So, um, And in, even with OCR technology, right, so there's Adobe APIs, there's
13:28
a lot of companies that do OCR, but one of the things we noticed was there's a
13:32
lot of inefficiencies with converting image to text with some of those APIs.
13:35
So we're actually using some of the latest technology there, right, of some
13:39
of these open source components, along with our proprietary kind of architecture
13:43
to be able to give you that best result.
13:46
So we're aiming for accuracy, right?
13:48
And making sure it's reliable and repeatability, and that's really
13:51
what our main focus is, right?
13:52
These are kind of things that are very, uh, almost an oxymoron with
13:58
AI, because that's really hard to do.
14:00
If you ask an AI the same question over and over again,
14:04
it'll give you different answers.
14:05
Like with chat GPT, if you ask it to generate an image of a person,
14:09
it'll generate that, but if you try to get it to regenerate the
14:11
same image, it's almost impossible.
14:13
Bob: Really?
14:14
Muneeb: Yes.
14:15
Bob: Well, I mean, that's, I mean, and the power of it too, I noticed
14:17
on your system, it wants to learn.
14:20
So it could tell you, you know, score my answer.
14:22
Why is it good?
14:23
Why is it bad?
14:23
Cause then it gets better.
14:24
It's learning, right?
14:25
Learning.
14:26
It's like, if you have, I had somebody, you know, say I use this platform
14:29
once a day, give me the right answer.
14:31
And I stopped using it.
14:32
And I was like, if a law clerk came to you and had the wrong
14:35
answer, would you fire them?
14:36
You wouldn't ask it like, let me do it this way.
14:38
Oh, and then it gets the right answer.
14:40
Right?
14:40
Right.
14:40
I mean, so what do you have to say to those lawyers out there that think.
14:43
That one, their data is not secure that my client's information has got leaked.
14:48
What's your answer to that for
14:48
Muneeb: CLIP?
14:49
So with that we have all the standards that the world provides at the moment
14:53
3 is coming as well So we're planning to do that as well to make you and
14:57
give you some of that comfort, right?
14:59
That you know your data is secure and it's private and we want you to feel that way
15:03
and if you read our privacy policy If you need the data, you know, we back it up.
15:08
Who owns
15:08
Bob: the data?
15:09
Muneeb: You own the data.
15:10
You, not the platform.
15:11
Not the platform.
15:11
Bob: That's, that's very powerful.
15:13
Muneeb: That's very powerful to me, especially like with me as well.
15:16
Like, uh, you know, that's one of the biggest challenges with us.
15:19
We're sharing so much information, our identities in the world.
15:22
We don't really own that data, right?
15:24
So that's one of the core concepts of the company as well is to make sure
15:28
that, you know, you're able to own your data because that's really, what's
15:31
going to be your currency in the future.
15:32
Bob: Yes.
15:32
Your own data and the better work product you have, the higher power,
15:36
most higher power data is the value.
15:38
Exactly.
15:39
And that's you as a law firm.
15:40
People always think about, well, like what's, You know, as a
15:43
law firm owner, sometimes you'd be like, well, what's my exit?
15:45
What am I able to do?
15:46
Well, if you have that amount of data and you have that much high level brain power
15:49
documents that could be queried and the answer could be very quick and the best,
15:53
Muneeb: well,
15:53
Bob: that's a different game.
15:54
Muneeb: That's right.
15:54
Bob: It's a different game.
15:55
Muneeb: That's right.
15:55
And that sets you apart from your competitors and also creates
15:58
better competition in the market.
16:00
Bob: Yeah.
16:00
So for those lower, and is there a way though, like that we could.
16:04
license the, the brain of our internal data.
16:08
So to other folks so that they can learn too.
16:10
Muneeb: Absolutely.
16:11
That's part of the vision of this, right?
16:12
So, uh, goal one right now, we just launched August 28th is to
16:16
get people on board, 2024, 2024.
16:20
Um, and you know, that's the, that's the game plan.
16:23
Once everyone's in there, uh, we'll give them a chance to monetize it.
16:26
We'll come up with marketplaces, different kinds of mechanisms to be
16:30
able to, Monetize and also get that return back right for for their data
16:34
as people are using it and sharing it
16:36
Bob: I mean imagine being able to query the best like bad faith
16:39
lawyer in the country Give him a case and say is this a case or not?
16:42
Why not?
16:43
And then oh man, I think it's closer than people think So SOC 2 compliance,
16:48
by the way, if you're if you're listening, it's S O C the number two.
16:51
That's right Uh is SOC 2 compliance We've talked about data security.
16:56
We've talked about querying chatbots.
16:58
That's right.
16:59
We've talked about not being replaced by artificial intelligence.
17:02
Muneeb: Yeah.
17:02
Bob: But being supplemented by it.
17:04
That's right.
17:04
Um, and what do you think overall, because I see so many companies popping up that
17:09
say they're artificial intelligence.
17:10
Yes.
17:10
What are some ways that people can fact check to make sure that it's real
17:15
and not just white label chat GBT?
17:17
Muneeb: Yeah, so, uh, there's, there's a couple of things, right?
17:20
One is the security, uh, one is the capability, one is the type of documents
17:25
and the size of documents you can actually upload and do, do things with.
17:29
Um, uh, those are, I would say, the three, you know, big things that,
17:33
you know, I would look at, right?
17:34
Especially security, especially Well,
17:36
Bob: security, you can ask them if they, if they're HIPAA compliant or SOC
17:39
2 compliant, because SOC 2 compliance incorporates HIPAA compliance.
17:42
Right.
17:43
But what about like the, what's a red flag if somebody says you can only download
17:47
like 10, 000 pages or something like that?
17:49
So
17:49
Muneeb: that tells you that, you know, if you can only download 10,
17:52
000 pages, then you know that they're capping it in the cloud, right?
17:56
And, and another thing is if people that actually use AI, it's very expensive.
17:59
It's a new technology.
18:01
So using these services in the cloud are the most expensive,
18:04
um, services right now.
18:06
Like, you know, actually having storage, compute, RAM are minuscule compared to.
18:10
What AI costs, right?
18:12
So when you're sending a prompt to one of these LLM models getting a response,
18:15
just that one interaction Each word costs money and that's what a token is, right?
18:20
So each Each of these things is costing money.
18:23
So if you're seeing like, for example, a low price, uh, either they have a
18:26
lot of money that they're going to burn, or it's something that, you know,
18:30
they're probably, let's talk about
18:32
Bob: price because people think this is incredibly expensive and things like this.
18:36
Clip is 500 bucks a month per user.
18:39
That's right.
18:39
That's that's insane.
18:41
It's a subscription.
18:42
It's not a pro run thing.
18:43
Muneeb: That's right It's a subscription and we want you to use it.
18:46
We want you to feel the value it if you don't like it You can cancel it, right?
18:50
That's really the value of this and and we came up with that price a lot of it
18:54
to do with not just a value because I think I always believe in 10x, right?
18:58
So if you're paying 500 bucks a month, you should be getting 5, 000 back with it.
19:02
If you're not, you should cancel it, right?
19:04
That's kind of how I, I feel 5, 000 or more of value, right?
19:09
So that's the philosophy here, right?
19:11
Is that's one factor.
19:12
The other factor is, Just the cost, right?
19:14
Based on the documents and things you're, you're bringing in.
19:17
The reason I'm not capping it, because everyone's gonna have a different
19:20
variance, and over time, we'll be able to balance that out, right?
19:24
Because some people will use it a lot, some people use it for few cases.
19:27
So I think over time, we'll be able to balance the cost, at the same
19:31
time, These massive tech companies are improving this every day and the costs
19:36
are going to come down in the cloud for, for some of these services, right?
19:39
And so, and so I think it's a fair price point and I think
19:44
based on the value you can get it.
19:45
And also with Clip, you know, one, one really interesting thing I learned
19:49
early in my career is like if you improve 1 percent every single day,
19:53
then by the end of the year you'll be 37 times better than you were today.
19:58
And this is the worst clip that math went way over my head Yeah, so but you
20:03
know, it's an interesting thing to think about right just 1 percent every single
20:07
day after a year You'll be 37 times better than you were if you actually
20:11
grew 1 percent every single day, right?
20:13
So so that to me is pretty Phenomenal and if you improve
20:17
Bob: 1 percent then you Degrade to the next day.
20:20
I mean, yeah, that's my struggle.
20:22
Yeah,
20:22
Muneeb: so well, then the math you'd have to subtract You'll
20:25
still be you can still balance it.
20:27
Yeah,
20:27
Bob: so, you know, thanks for coming on the show.
20:29
How do people find you?
20:30
Muneeb: Yeah, so they can go to our website Kiowa dot AI and then
20:34
sign up for clip and get started.
20:36
Bob: So what's could you spell that for our listeners, please?
20:38
Muneeb: Yes It's K I o W a dot
20:42
Bob: AI Perfect.
20:44
And you know, we could probably work on getting our justice team
20:47
podcast listeners a discount to clip.
20:48
Maybe we can put a discount code on their 10 percent off.
20:51
What do you think on the spot?
20:52
I think so.
20:53
I love it.
20:54
So we'll drop that into the notes.
20:56
Um, wherever you're listening, whether it's YouTube or in the apple
20:59
channel, something like that, we'll drop in a discount code for clip.
21:02
And so far we've been up and running at our firm with several
21:04
users and we love using it.
21:06
We look forward to, you know, enhancing more, uh, Our capabilities
21:10
and making us more efficient because on our side of the coin, we get
21:13
paid perhaps not pay it per hour.
21:14
So we gotta be real efficient.
21:16
Muneeb: That's right.
21:16
Bob: Um, thank you for coming on the show.
21:19
Muneeb: Thank you so much, Bob, for having me.
21:20
And thanks everyone.
21:21
Bob: Thank you for listening to this episode of the justice team
21:24
podcast on the justice team network.
21:26
Reach out.
21:26
If you have any questions to justice team network.
21:28
com or if you want to get in touch with clip, we'd be happy
21:31
to make the introduction.
21:32
Thanks everyone.