June 27, 2025 • 31 mins
Everyone is talking about building AI agents. But is your foundation even ready?
In this episode, Andreas Welsch speaks with Barr Moses, CEO of Monte Carlo, about the hidden risks behind the push for Agentic AI and what business leaders are missing when they rush to scale.
Barr shares what she’s hearing from data and AI leaders: The majority is building or deploying AI this year, but only 1 in 3 believe their data is ready, and even few have a reliable way to ensure agent outputs are correct.
What does that mean for your AI roadmap? A strong model or prompt isn’t enough. If the data is wrong, the agent’s action will be wrong and in some cases, costly. From AI chatbots selling cars for $1 to data platforms returning flawed recommendations, the risks are real.
Barr also breaks down three areas data and AI teams must prioritize:
- Productivity by using AI to accelerate their own workflows
- Readiness to build a foundation of high-quality, trustworthy data
- Realiability by ensuring AI agents produce outputs that align with business goals
If you’re working on an AI strategy or are leading teams expected to deliver on it, this conversation makes the case for why AI success is built upon operational readiness.
Don't miss out on the conversation – tune in now to learn how to turn AI hype into business outcomes.
Questions or suggestions? Send me a Text Message.
***********
Disclaimer: Views are the participants’ own and do not represent those of any participant’s past, present, or future employers. Participation in this event is independent of any potential business relationship (past, present, or future) between the participants or between their employers.
Level up your AI Leadership game with the AI Leadership Handbook:
https://www.aileadershiphandbook.com
More details:
https://www.intelligence-briefing.com
All episodes:
https://www.intelligence-briefing.com/podcast
Get a weekly thought-provoking post in your inbox:
https://www.intelligence-briefing.com/newsletter
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Andreas Welsch (00:00):
Today we'll talk about getting AI ready with
reliable data, and who better totalk about it than someone who's
actively working on that bar.
Moses, Hey Barr.
Thank you so much for joining.
Barr Moses (00:09):
Thanks for having me.
Andreas Welsch (00:11):
Wonderful.
Why don't you tell our audiencea little bit about yourself, who
you are and what you do.
Barr Moses (00:16):
Yeah, happy to.
As mentioned, my name is BarrMoses.
I'm the CEO and co-founder of acompany called Monte Carlo.
Monte Carlo is the leader andcreator of a category called
Data and AI Observability.
And what we do is we work withhundreds of enterprises ranging
from companies like Cisco andIntuit to leading.
(00:36):
Airlines, like American Airlinesto leading retail and CPG like
PepsiCo and many otherorganizations.
And the thing that's common toall of them is they use data and
AI to power their business tobuild better experiences for
their customers to build better.
Solutions for their customers,all powered by data, and AI.
(00:59):
And what Monte Carlo does ispartner with these companies to
help them make sure that theirdata and AI products are
actually reliable and can betrusted.
The worst thing is when they're,when you look at you steer at a,
an agent gives you a totallywrong answer that you totally
did not expect.
Or maybe you're looking at areport that's based totally on
the wrong numbers.
(01:20):
In those instances, you gotta.
Lose trust and the, data and AIteam's credibility is on the
line.
Their brand, their reputation ison the line.
And so we help data and AI teamsknow about issues before they
happen and be able to catch thembefore they have catastrophic
impact on businesses.
Andreas Welsch (01:39):
That's wonderful.
Thank you so much for, sharingthat.
And I think, when, we're lookingat leaders being asked to go
figure out AI, go exploreagents.
I think a lot of times that'sjust a disaster waiting to
happen because you are not asinformed maybe, or you are just
doing this for the first time.
(02:00):
So you think Yeah it's maybe aseasy as some of the vendors
advertising these platforms makeit sound.
But at the end of the day, itdoes come back to your own data
and to your company's data.
So really excited about theconversation with you today.
Should we play a little game tokick things off?
What do you think?
Barr Moses (02:18):
Sure.
I'm not a very fun person.
I have to warn you of that, buthappy to see.
I'm happy to participate.
That's fine.
Andreas Welsch (02:26):
We'll see.
So this one is called In yourown Words, I'd like for you to
answer with the first thing thatcomes to mind and why.
And for you in the audience, I'dlove to get your answer as well
to make it a little moreinteresting.
You only have 60 seconds foryour answer.
So are you ready for, What's theBUZZ?
Barr Moses (02:44):
Born ready.
Andreas Welsch (02:45):
Okay, good.
So let's see.
If AI were a book, what would itbe?
Barr Moses (02:53):
Oh, great question.
Andreas Welsch (02:54):
60 seconds on the clock.
Barr Moses (02:57):
That is a great question.
If AI were a book, what would itbe?
Trying to think of a goodblockbuster.
One of the books that I've readrecently that somewhat related,
but not entirely is a book byformer Snowflake CEO.
(03:18):
His name is Frank Slootman andthe book is called Amp It Up.
I was just remembered of thatbook.
And the whole sort of premise ofthe book is around how can you
amp it up and by that meansdrive up the intensity and the
urgency as a culture.
And I think if AI has donesomething, it's definitely amped
(03:38):
it up for everyone.
Everyone is there's an increasedurgency, increased intensity,
increased innovation cycles,increased products have, are
changing on a weekly and monthlybasis.
New foundation models are beingreleased left and right.
And so everyone really has toamp it up.
(03:58):
Everyone has a harder job.
That's what come, that's whatcomes to mind.
Andreas Welsch (04:03):
I love that.
Great answer.
And indeed it's a lot aboutamping amping it up.
I saw a poll yesterday by Axioswhere they have Americans across
all different demographics andthe sentiment was yeah, yes,
it's great to have so much AI,but actually can we slow it down
(04:24):
a little bit?
Can we actually tone it down andbe more deliberate?
So it's interesting to see howthese different demand dynamics
play out.
And I think a lot of people havealso have have had access to
generative AI tools and havebeen able to see this.
And there's obviously been a lotof discourse in the media and so
on.
So interesting to see how thatperception shifts as well.
(04:46):
But definitely right, we need toact with urgency, but we also
need to make sure that we acton, the right things.
A couple weeks ago I was meetingwith the C-suite leaders on an
engineering company and, theyshared their vision for how AI
should help them operate thebusiness better.
And throughout that conversationwe actually realized that they
(05:06):
first need to have a datafoundation in place which they
don't have yet.
It's actually not so much aboutan AI transformation, but more
about a business transformation.
We'd love to do win-lossanalysis.
Great.
How are you capturing the data?
In people's minds or onspreadsheets, maybe if we're
lucky.
So we need to ask people.
Okay.
So there are some fundamentalthings and I'm wondering what
(05:29):
are you seeing in yourdiscussions?
Certainly the companies youmentioned are well underway on
their AI and machine learningand Gen AI journey.
But I'm sure you're meeting withothers as well.
So are engineering companieslike the one that I met with,
the only ones that are facingthis or are others are going
through similar challenges too?
Barr Moses (05:48):
Yeah, I think it's a great question.
Maybe a reaction to that andalso what you said prior.
I think let's just start byfacing, I think AI facing
reality.
I think the, pressure around AIis real.
I.
And that's happening everywherein boardrooms and in the media
and with your peer groups.
We actually just did a, surveywith a couple hundred data and
(06:11):
AI leaders and asked them anumber of questions about their
approach to data foundations,AI, and the results were pretty
striking.
A hundred percent of leaders.
Are currently have AI inproduction or are planning on
having AI in production thisyear?
So clearly everyone is feelingthe hype and everyone is acting
(06:33):
on it, right?
I don't think that's new at all.
However, only two, only one outof three respondents actually
think their data is ready forAI.
Now, in a world where we live,where a hundred percent of
people are working on AI, butthe large majority of people
think that their data is notready in AI, we're obviously
faced with.
A problem, if you will.
(06:54):
Now I think what does being AIready mean for your data?
We can unpack what that means,but maybe just to double click
into the implications here formany organizations, I was just
talking to CTO of a Fortune 100company, and basically he told
me like, look, by the end ofthis year, I expect us to have
(07:16):
over 500 agents.
That out in the wild that we'vebuilt.
And agents are not deterministicsystems.
And so I cannot for certaintypredict what the output of a
model can be.
And so I might have 500 agentsout in the wild sharing
unpredictable outputs that Ihave no oversight on.
(07:38):
And so that reality is a scaryreality for people.
Yes.
And the the implications onrevenue and brand and risk are
real.
Just to give you a coupleexamples, a couple years ago
Citibank was actually hit with aseveral hundred million dollars
for data quality issues.
And so you can actually, there,there's severe regulatory risks
(08:01):
for.
For practices that are notstrong enough in data quality
that's been here for, a while.
And that's not new in the lastfive to 10 years has been
multiple recent issuesdescribing catastrophic results
and impact of the data beingdata AI platform being wrong.
(08:21):
Fast forward to today.
Just a couple of months ago auser actually was able to
convince a chatbot, a ChevyTahoe car to to, purchase that
car for$1.
An agent, basically, again, anagent sold a Chevy Tahoe car for
(08:42):
$1 because the user was able toconvince the chat bot to do
that.
And obviously you can imaginethe repercussions.
For, in that instance.
And so organizations far andwide, on the one hand, need to
invest a ton in AI or beingasked, being tasked.
And AI, on the other hand, ahundred percent with Trinity, I
(09:03):
can tell you the foundations arenot there.
And the ability to deliverreliable data and AI products is
becoming more important thanever in this world.
Andreas Welsch (09:14):
So I, was just at Data Summit in Boston about
two weeks ago where lots of dataleaders met from data
management, data governance,many different roles, heads of
analytics and so on.
And they shared a similar thing,but they also shared the concern
that I'm not getting the budgetfor it.
My boss says, we have lots ofbudget that we can assign to AI
projects.
Can you come up with an idea foran AI project?
(09:36):
But I know that the data isn'twhat it needs to be.
It's not complete, it's notaccurate, it's not fresh and
what have you.
What are you seeing there?
How are leaders coping withthat?
That yes, on one hand there,there is budget, but management
ask me to invest it in shinythings.
That we can, again, report upand look at and say, we're doing
(09:56):
AI, but we actually know that weneed to fix the foundation
before we can do something thatis reliable, that will not
expose us to a significant risk.
Barr Moses (10:05):
Yeah, it's a good question.
I I think by and large from whatwe can see, most companies have
budgets and are willing toinvest in AI.
A lot of it is experimentalbudgets or sort of innovation
budgets.
And so I think there's aquestion of what will be sticky
and what will be here in 18 to24 months.
(10:26):
Sounds like that resonates.
I think that's certainly timewill tell, but I do I think it's
hard.
It's very rare that I comeacross companies that don't
invest in AI.
Now when I think about data andAI teams, there's primarily
three core problems that I heardata and AI teams are faced
(10:48):
with, and I think thatinfluences their budget decision
making.
So I'll just walk through whatthese three core problems are.
The first core problem is thatthey are being tasked just like
every other team, just likeevery other function, they're
being tasked with finding waysto accelerate the output and the
productivity.
Of their team with AI andautomation.
(11:10):
So every single team,engineering support, customer
success data and AI teams,they're being asked to do more
with AI.
And so that means things thatwere workflows that we're
already doing, things that we'realready spending time on doing
them faster and better using AI.
That's like the first kind ofcore problem that they have.
The second core problem thatdata and AI leaders face is.
(11:34):
They are, they own a dataplatform and they own a lot of
enterprise data, a lot ofproprietary data.
And that data is feeding AIproducts.
Maybe it could be a chat bot, itcould be an agent, whatever it
is.
I.
And the data that they'reproviding to these AI solutions
needs to be reliable.
Now, why does this data evenmatter?
(11:56):
Because when I'm building an AIproduct, everyone has access to
the latest and greatest model.
I can always switch betweenAnthropic and open AI and
something else.
We all have access to that withjust a few clicks.
And I have an API and I'm done.
But the thing that I have thatmy competitor or another,
company doesn't have is I haveproprietary data.
(12:16):
I have enterprise data, so Iknow I have more information and
can actually tailor theseproducts to offer a better
customer experience.
I'll give you an example.
If you work with an airlinecompany for example, or if you
work with a hotel chain I canoffer a recommendation product.
For example, Hey maybe youwanna.
Have this for lunch or have thisexperience at the hotel, I can
(12:39):
offer a better solution or amore personalized concierge, if
you will, because I know yourpreferences and I have prior
information about what's yourlunch preferences or whatnot.
And so I can actually, I.
Make more intelligencerecommendations for you.
If that data is unreliable, ifI'm using the wrong data to feed
those AI products, thenobviously everything crumbles,
(13:01):
right?
So that's the second kind ofcore issue.
Make my data AI ready.
And a lot goes into that.
We can go into more detail.
There's you have to make surethat you have structured data is
reliable, your unstructured datais reliable.
That this whole thing is like abig thing, right?
But, that's the, second coreissue.
(13:21):
Then the third core issue that Ireally see data and AI leaders,
struggle with or find the needto divert attention and
resources to is now that they'vereleased AI products, how do we
make sure that those arereliable?
So how do we make sure that theagent is not selling the Chevy
Tahoe car for$1?
(13:41):
Or because you can have theperfect trainee data the perfect
prompt, the perfect context, butthe output of the model will
still not be fit for pur forpurpose.
And again.
Can go into more detail and sothat, but at a very high level
when I think about decisionmaking for a company and for,
budgets, these three things arehonestly table stakes.
(14:04):
I would say the first one,certainly like making data teams
more productive.
Yes.
With AI and automation.
Really table stakes.
I think the second more and moretable stakes, like getting the
foundation ready, making, havingAI ready data.
I think the third category,building reliable AI solutions
is probably an area that peopleare still in the early days of.
We're still many companies arejust moving to the cloud, right?
(14:29):
They're still in that, on thatjourney and there's a lot that
goes into actually delivering AIsolutions.
But at a high level, these arelike the three, three big
problems that the data and AIleaders have to tackle.
Andreas Welsch (14:41):
I think that makes it very tangible, right?
And also shows this progressionof how, should you think about
this?
Also in, in three simple stepsnow.
You mentioned yes we, need toput more emphasis on, data.
We need to empower data teamsmore.
What are you seeing, how areleaders getting buy-in along the
chain of command to do thesedata projects?
(15:01):
Again, when everybody's askingwhat's our AI strategy?
Barr Moses (15:04):
Yeah, it's a good question.
Look I think one of the thingsthat is really important and
obvious, but like easier saidthan done is tying what this is
to.
Tying what you're building to,to real value and to real impact
and maybe even just saving, cofinding cost savings along the
way using AI.
(15:25):
And by that lemme just give youlike a practical example of how
you can use AI to save time.
At Monte Carlo, what I mentionedwe offer observability for
companies or help companies makesure that their data and AI
products or their data and AIestate is reliable.
And we also use AI ourselves.
(15:45):
And so we actually built or weare building a number of agents.
We're building an observa suiteof observability agents.
And the goal of this agent isbasically to deliver real hard
ROI for data and AI teams.
And, I'll give you an example orkind of walk you through how
these works both'cause I thinkit's super cool.
(16:06):
But also because I think, Iactually think it's valuable for
data and AI team.
One of the things that datateams spend a lot of time on,
mostly data analysts spend a tonof time on, is trying to figure
out what in their data needs tobe monitored.
So I have a lot of tablessometimes thousands, tens of
thousands, hundreds of thousandsof tables in my.
(16:27):
Data lakehouse into my datawarehouse.
That could be at your GCPplatform, Azure, Databricks,
Snowflake, AWS, what have you.
And if I'm a data analyst or adata engineer, I need to make
sure that data is accurate, butI have no idea.
I.
What are their requirements?
It's hard for me to be able tospecify at the table or even
(16:48):
feel level to know exactly whatneeds to be accurate, what the,
definition of accuracy evenmeans.
And so oftentimes what datateams do is they spend time
rigorously and tediously, firstof all, profiling the data.
So like understanding thestructure of the data and
understanding the data itself.
Digging through that, lookingthrough metadata, making
(17:10):
connections to understand thesemantic meaning of fields.
And then coming up with specificmonitors that might say, oh, you
know this, I'm just making thisup.
Again.
If this is like a, an airlinethe, and I have a column that
has number of every row is aflight.
(17:30):
And so I need to make sure thatthe flight numbers have a
certain sort of they need tolook a certain way, they need to
have a certain number ofcharacters, et cetera.
So I need to have a monitor tomake sure that that data is
always accurate.
'cause I can't mess up flightnumbers, right?
Yes.
Like flight numbers, data needto be accurate.
And so that process is really,hard and tedious and manual to
(17:54):
go through all of that and comeup with monitors.
What we have done is actuallyreleased a monitoring agent.
It's in production already.
It's used by hundreds ofcustomers, so it's been out live
and for, a while.
We actually have a 60%acceptance rate, which means
that 60% of the monitors that werecommend are being used and get
accepted.
And what we do is actually, thisagent actually weird to say
(18:18):
this, but it mimics the humanbehavior.
It goes through, it profiles thedata, it tries to make this the
understand the meaning ofdifferent connections between
the data and then makerecommendations for what you
need to monitor.
And so that cuts down for a datateam time from like weeks to
(18:39):
minutes if you will.
And so it's really cool to startseeing what you can actually
start offering to data and AIteams to make them more
productive.
And so what we're building andwe're about to release in the
next few weeks.
And this is what I think is areally cool application of LLMs
has to do with troubleshooting.
So the first thing that data andAI teams do is they monitor
(19:01):
data, right?
They need to know when thedata's wrong.
The second thing that they do iswhen the data is wrong, they
need to start.
They have a workflow forunderstanding why it's wrong,
starting to triage andtroubleshoot and try to
understand what was the rootcause.
And so what our, troubleshootingagent does is again, mimics what
a data gov data steward woulddo.
(19:22):
What a data steward data analystwould typically do is they would
start with coming up withhypothesis.
For what might be wrong.
So let's say I get anotification that there's a
particular issue with aparticular report.
And this report is being usedevery single day by our field
operations team.
It's high visibility.
(19:42):
It's really important that I.
Look into this issue.
It's, data.
It's a report that's being usedall the time, and then I start
coming up with hypothesis forwhat could go wrong.
And I start going upstream tableby table and saying, okay, let
me check if the data arrived ontime.
Okay, let me check if thisupstream data arrived on time.
Okay, now let me check thatanyone make a change to the
(20:03):
code.
Somewhere that anyone breaksomething.
Okay.
Maybe the ETL system, maybe thethe job failed and then
incomplete.
I basically had to come up witha list of dozens of hypothesis
and start to check them and testand cover what happened.
What we've done in, thistroubleshooting agent is we
basically have an ensemble ofLLMs.
(20:24):
There's an l there's this masterLLM that comes up with the list
of hypothesis, and then itspawns off a new agent for every
single hypothesis.
And so every agent thenbasically recursively looks into
a particular hypothesis, and soyou can have up to, I think
around a hundred or so agentsrunning in less than two
(20:45):
minutes.
All looking into differenthypothesis for what could go
wrong at the same time.
And so something that could havetaken me years frankly, to go
through all of this hypothesisand try to understand what goes
wrong with a troubleshootingagent, again, with a combination
of.
Breaking down the problem intodifferent tasks.
Every task is a hypothesis tolook into.
(21:08):
And then using a lot of LLMs tolook into each of this
hypothesis and then basicallycome back and report back on
what they found.
And then this sort of master,there's a, there's sort of a.
Bigger LLM that kind ofsummarizes all that and gives
you like A-T-L-D-R, here's theroot cause what happened this is
the reason for this incident isX, Y, Z.
(21:29):
And so that is really powerful.
So you're basically taking,again, things that data and AI
teams are already doing, likethey're already spending time on
that basically cutting theamount of time significantly.
And so the ROI in thoseinstances is very clear, if that
makes sense.
Andreas Welsch (21:46):
Great example.
And very tangible too.
I love these examples where it'sabout, hey, it would have taken
one person or a team of peoplemonths or years to do this.
Now we can do this in minutes oreven less than a couple minutes.
So super powerful, and I thinkwe need more of these examples
that show what can this actuallydo.
(22:07):
Beyond the hype talk and totallythese kind of things.
So sounds like you are alreadyworking on some exciting things
and are testing this withcustomers or rolling this out to
customers.
Super, super cool.
Now, it's been a few weeks.
I think I would say it'sprobably even been a few months
(22:27):
since we've started talking andsetting up today's session.
And I remember probably earlierin the year I came across an
article from you where it saidAgentic AI might actually be
doomed to fail.
And we, somehow landed on thistopic of getting data AI ready.
But I'm wondering why do youthink AI agents might be doomed
(22:48):
to fail?
When are they doomed to fail?
It's probably the betterquestion to ask.
When are they doomed
Barr Moses (22:53):
to fail?
Yeah, good question.
Look we talked about thisearlier on this call when folks
are releasing hundreds of agentsinto the wild and like waiting
to see what happens.
I think we know what happens.
I think we know what the storylooks like here.
And I love this example of sortof the troubleshooting agent
(23:14):
because I just wanna stay, Ijust wanna say when obviously
when our team presented to me,shared we were working on the
troubleshooting agent.
I think that was like a turningpoint for me because I think,
there's a lot of AI skepticismout there and there's a lot of
question of what does itactually mean?
(23:34):
And just to be clear I don'tthink AI will replace people or
replace data and AI teams.
If anything if I look at acorollary like engineering, as
engineering as a space, the moreadvancements we've had with
engineers the higher there was ademand for engineering teams.
And I think the same will be fordata and AI teams.
And the more we use AI, the morewe need data and data, there's
(23:57):
gonna be an increase in demand.
It doesn't mean that we don'tneed to change how we work and,
what we do.
And I'm, I very much believe inthat.
And also I think thisapplication of the
troubleshooting agent reallybrought to life for me What.
How powerful if you put LLMs touse, how powerful that could be.
(24:20):
And so you can imagine a lot ofcomplex problems that you might
be working on today that if wecan intelligently break them
into smaller tasks and usedifferent, smaller and larger
LLMs in different instances, youcould actually teach LLMs to do
really, clever things.
And so I wanna start I, thinkit's important me to say that
because I think.
(24:41):
I think figuring out what agentscan do for your business is
important.
And I do believe, I am convincedtoday that there is value in
that for your company, for yourbusiness.
And I think you have to see itand to experience it yourself,
to believe it.
It's very different when it's intheory.
(25:03):
So first of all I and, there's alot of talk about like, how in,
in a couple years, softwareengineers will really just be
managing a fleet of agents.
Like you'll have one agent tobuild features, one agent to fix
your bugs, one agent to do PRDswith you basically have a fleet
of agents.
And so similarly you canenvision that in the world, data
(25:24):
and AI teams have like a fleetof agents one agent to monitor
their data, one agent totroubleshoot their, and we'll
all just sit here in our cushychair and monitor all the
agents.
I can't wait for that.
I obviously like that means thatour world will, be very
different and how we work willbe very different, but, in that
(25:45):
world, or the reason why I wrotethat, that I think there's a
high chance that AgTechsolutions will fail is if we,
are not thoughtful about what itmeans to build high trust
agents.
And I'll give you an example,the the most like popular
(26:09):
example, if you will think thiswent viral on XA couple years
ago.
Or, maybe a couple last year, Ithink someone wrote on, on one
of sort of Google's LLMsolutions.
What should I do if cheese isslipping off my pizza?
Yes, you may have seen this andthe answer was, oh, you should
(26:30):
just use organic super glue.
No problem.
To like glue the cheese backonto your pizza.
Now, just to be clear, like thetraining data was good.
The like prompt was good.
The con, everything was good,right?
I.
But the output was totallyinappropriate and made no sense
for the context.
Now I will continue to useGoogle or maybe use perplexity
or something like, but Googlecan get away with it.
(26:53):
We can't get away with that.
Most organizations in the worldcan't get away with building
agents that will have, that willshare responses like that.
That are really inappropriate.
And, continue to have trust.
And so I think the way that weneed to manage our agents and
(27:14):
the way that we need to manageour data and AI platform needs
to change drastically.
And one of the things that we'vedone at Monte Carlo over the
last several years is work withthousands of enterprise and help
figure out.
What is the way to buildreliable data and AI products
(27:34):
and based on, again, thousandsof, customers and conversations,
we've found that problems withdata and AI products can really
be reduced to four coreproblems.
The first is data and AIsolutions can be running on bad
data, meaning the data thatyou're feeding, it might be
inaccurate or late or justwrong.
(27:56):
The second core reason for whydata and AI systems might be
wrong is because there's a codechange.
Maybe it's a bad join maybe it'sa schema change.
Maybe it's a code change in theagent itself that's actually
changed something dramatically.
So the third reason for whythings can go wrong is if
(28:18):
there's a problem with thesystem.
A system that could be like yourETL jobs, like airflow or DBT,
or it could be orchestrators, itcould be things like land chain
or land graph.
Basically anything along yourdata and AI say can go wrong in
your system.
'cause the one thing for sure, ahundred percent of systems fail
at some point.
(28:39):
And then the fourth thing thatcan go wrong, and I alluded to
this earlier, is.
You can have all those thingswork perfectly.
You can have the perfectcontext, the perfect prompt, but
still the map model output willbe not fit for purpose.
And so I think for people whoare building agents who are
building AI solutions, unless westart thinking about the
(29:00):
holistic health of those systemsoverall as a function of those
four things, data code systemsand model output, we will fail.
Andreas Welsch (29:11):
Wow, that's that really hits home and I think
comes down to the core of itreally.
Barr, thank you so much forsharing this.
We've covered a lot of ground inthe last 30 minutes.
From helping data teams empowerthem and understand that they
have big responsibility in whatthey can do to observability,
(29:32):
making sure that our data isactually accurate and correct.
And then talk about some of therisks, what happens when you
don't do that, and where thefailure points in, Agentic AI.
Also, great to hear what you'redoing at Monte Carlo and how
you're bringing AI and agents todata and AI teams.
Sounds like a really excitingopportunity and really exciting
space too.
Barr Moses (29:50):
So yeah.
Thanks.
Andreas Welsch (29:53):
Yeah, wonderful.
Thank you so much for joining usand for sharing your experience
with us today.
Barr Moses (29:57):
Absolutely.
Thank you for having me.
It's an exciting time.
There's a lot of exciting stuffhappening, I think the future is
bright.
I'm excited to be part of it.
And thank you to everyonejoining.
Today we'll talk about getting AI ready with
reliable data, and who better totalk about it than someone who's
actively working on that bar.
Moses, Hey Barr.
Thank you so much for joining.
Barr Moses (00:09):
Thanks for having me.
Andreas Welsch (00:11):
Wonderful.
Why don't you tell our audiencea little bit about yourself, who
you are and what you do.
Barr Moses (00:16):
Yeah, happy to.
As mentioned, my name is BarrMoses.
I'm the CEO and co-founder of acompany called Monte Carlo.
Monte Carlo is the leader andcreator of a category called
Data and AI Observability.
And what we do is we work withhundreds of enterprises ranging
from companies like Cisco andIntuit to leading.
(00:36):
Airlines, like American Airlinesto leading retail and CPG like
PepsiCo and many otherorganizations.
And the thing that's common toall of them is they use data and
AI to power their business tobuild better experiences for
their customers to build better.
Solutions for their customers,all powered by data, and AI.
(00:59):
And what Monte Carlo does ispartner with these companies to
help them make sure that theirdata and AI products are
actually reliable and can betrusted.
The worst thing is when they're,when you look at you steer at a,
an agent gives you a totallywrong answer that you totally
did not expect.
Or maybe you're looking at areport that's based totally on
the wrong numbers.
(01:20):
In those instances, you gotta.
Lose trust and the, data and AIteam's credibility is on the
line.
Their brand, their reputation ison the line.
And so we help data and AI teamsknow about issues before they
happen and be able to catch thembefore they have catastrophic
impact on businesses.
Andreas Welsch (01:39):
That's wonderful.
Thank you so much for, sharingthat.
And I think, when, we're lookingat leaders being asked to go
figure out AI, go exploreagents.
I think a lot of times that'sjust a disaster waiting to
happen because you are not asinformed maybe, or you are just
doing this for the first time.
(02:00):
So you think Yeah it's maybe aseasy as some of the vendors
advertising these platforms makeit sound.
But at the end of the day, itdoes come back to your own data
and to your company's data.
So really excited about theconversation with you today.
Should we play a little game tokick things off?
What do you think?
Barr Moses (02:18):
Sure.
I'm not a very fun person.
I have to warn you of that, buthappy to see.
I'm happy to participate.
That's fine.
Andreas Welsch (02:26):
We'll see.
So this one is called In yourown Words, I'd like for you to
answer with the first thing thatcomes to mind and why.
And for you in the audience, I'dlove to get your answer as well
to make it a little moreinteresting.
You only have 60 seconds foryour answer.
So are you ready for, What's theBUZZ?
Barr Moses (02:44):
Born ready.
Andreas Welsch (02:45):
Okay, good.
So let's see.
If AI were a book, what would itbe?
Barr Moses (02:53):
Oh, great question.
Andreas Welsch (02:54):
60 seconds on the clock.
Barr Moses (02:57):
That is a great question.
If AI were a book, what would itbe?
Trying to think of a goodblockbuster.
One of the books that I've readrecently that somewhat related,
but not entirely is a book byformer Snowflake CEO.
(03:18):
His name is Frank Slootman andthe book is called Amp It Up.
I was just remembered of thatbook.
And the whole sort of premise ofthe book is around how can you
amp it up and by that meansdrive up the intensity and the
urgency as a culture.
And I think if AI has donesomething, it's definitely amped
(03:38):
it up for everyone.
Everyone is there's an increasedurgency, increased intensity,
increased innovation cycles,increased products have, are
changing on a weekly and monthlybasis.
New foundation models are beingreleased left and right.
And so everyone really has toamp it up.
(03:58):
Everyone has a harder job.
That's what come, that's whatcomes to mind.
Andreas Welsch (04:03):
I love that.
Great answer.
And indeed it's a lot aboutamping amping it up.
I saw a poll yesterday by Axioswhere they have Americans across
all different demographics andthe sentiment was yeah, yes,
it's great to have so much AI,but actually can we slow it down
(04:24):
a little bit?
Can we actually tone it down andbe more deliberate?
So it's interesting to see howthese different demand dynamics
play out.
And I think a lot of people havealso have have had access to
generative AI tools and havebeen able to see this.
And there's obviously been a lotof discourse in the media and so
on.
So interesting to see how thatperception shifts as well.
(04:46):
But definitely right, we need toact with urgency, but we also
need to make sure that we acton, the right things.
A couple weeks ago I was meetingwith the C-suite leaders on an
engineering company and, theyshared their vision for how AI
should help them operate thebusiness better.
And throughout that conversationwe actually realized that they
(05:06):
first need to have a datafoundation in place which they
don't have yet.
It's actually not so much aboutan AI transformation, but more
about a business transformation.
We'd love to do win-lossanalysis.
Great.
How are you capturing the data?
In people's minds or onspreadsheets, maybe if we're
lucky.
So we need to ask people.
Okay.
So there are some fundamentalthings and I'm wondering what
(05:29):
are you seeing in yourdiscussions?
Certainly the companies youmentioned are well underway on
their AI and machine learningand Gen AI journey.
But I'm sure you're meeting withothers as well.
So are engineering companieslike the one that I met with,
the only ones that are facingthis or are others are going
through similar challenges too?
Barr Moses (05:48):
Yeah, I think it's a great question.
Maybe a reaction to that andalso what you said prior.
I think let's just start byfacing, I think AI facing
reality.
I think the, pressure around AIis real.
I.
And that's happening everywherein boardrooms and in the media
and with your peer groups.
We actually just did a, surveywith a couple hundred data and
(06:11):
AI leaders and asked them anumber of questions about their
approach to data foundations,AI, and the results were pretty
striking.
A hundred percent of leaders.
Are currently have AI inproduction or are planning on
having AI in production thisyear?
So clearly everyone is feelingthe hype and everyone is acting
(06:33):
on it, right?
I don't think that's new at all.
However, only two, only one outof three respondents actually
think their data is ready forAI.
Now, in a world where we live,where a hundred percent of
people are working on AI, butthe large majority of people
think that their data is notready in AI, we're obviously
faced with.
A problem, if you will.
(06:54):
Now I think what does being AIready mean for your data?
We can unpack what that means,but maybe just to double click
into the implications here formany organizations, I was just
talking to CTO of a Fortune 100company, and basically he told
me like, look, by the end ofthis year, I expect us to have
(07:16):
over 500 agents.
That out in the wild that we'vebuilt.
And agents are not deterministicsystems.
And so I cannot for certaintypredict what the output of a
model can be.
And so I might have 500 agentsout in the wild sharing
unpredictable outputs that Ihave no oversight on.
(07:38):
And so that reality is a scaryreality for people.
Yes.
And the the implications onrevenue and brand and risk are
real.
Just to give you a coupleexamples, a couple years ago
Citibank was actually hit with aseveral hundred million dollars
for data quality issues.
And so you can actually, there,there's severe regulatory risks
(08:01):
for.
For practices that are notstrong enough in data quality
that's been here for, a while.
And that's not new in the lastfive to 10 years has been
multiple recent issuesdescribing catastrophic results
and impact of the data beingdata AI platform being wrong.
(08:21):
Fast forward to today.
Just a couple of months ago auser actually was able to
convince a chatbot, a ChevyTahoe car to to, purchase that
car for$1.
An agent, basically, again, anagent sold a Chevy Tahoe car for
(08:42):
$1 because the user was able toconvince the chat bot to do
that.
And obviously you can imaginethe repercussions.
For, in that instance.
And so organizations far andwide, on the one hand, need to
invest a ton in AI or beingasked, being tasked.
And AI, on the other hand, ahundred percent with Trinity, I
(09:03):
can tell you the foundations arenot there.
And the ability to deliverreliable data and AI products is
becoming more important thanever in this world.
Andreas Welsch (09:14):
So I, was just at Data Summit in Boston about
two weeks ago where lots of dataleaders met from data
management, data governance,many different roles, heads of
analytics and so on.
And they shared a similar thing,but they also shared the concern
that I'm not getting the budgetfor it.
My boss says, we have lots ofbudget that we can assign to AI
projects.
Can you come up with an idea foran AI project?
(09:36):
But I know that the data isn'twhat it needs to be.
It's not complete, it's notaccurate, it's not fresh and
what have you.
What are you seeing there?
How are leaders coping withthat?
That yes, on one hand there,there is budget, but management
ask me to invest it in shinythings.
That we can, again, report upand look at and say, we're doing
(09:56):
AI, but we actually know that weneed to fix the foundation
before we can do something thatis reliable, that will not
expose us to a significant risk.
Barr Moses (10:05):
Yeah, it's a good question.
I I think by and large from whatwe can see, most companies have
budgets and are willing toinvest in AI.
A lot of it is experimentalbudgets or sort of innovation
budgets.
And so I think there's aquestion of what will be sticky
and what will be here in 18 to24 months.
(10:26):
Sounds like that resonates.
I think that's certainly timewill tell, but I do I think it's
hard.
It's very rare that I comeacross companies that don't
invest in AI.
Now when I think about data andAI teams, there's primarily
three core problems that I heardata and AI teams are faced
(10:48):
with, and I think thatinfluences their budget decision
making.
So I'll just walk through whatthese three core problems are.
The first core problem is thatthey are being tasked just like
every other team, just likeevery other function, they're
being tasked with finding waysto accelerate the output and the
productivity.
Of their team with AI andautomation.
(11:10):
So every single team,engineering support, customer
success data and AI teams,they're being asked to do more
with AI.
And so that means things thatwere workflows that we're
already doing, things that we'realready spending time on doing
them faster and better using AI.
That's like the first kind ofcore problem that they have.
The second core problem thatdata and AI leaders face is.
(11:34):
They are, they own a dataplatform and they own a lot of
enterprise data, a lot ofproprietary data.
And that data is feeding AIproducts.
Maybe it could be a chat bot, itcould be an agent, whatever it
is.
I.
And the data that they'reproviding to these AI solutions
needs to be reliable.
Now, why does this data evenmatter?
(11:56):
Because when I'm building an AIproduct, everyone has access to
the latest and greatest model.
I can always switch betweenAnthropic and open AI and
something else.
We all have access to that withjust a few clicks.
And I have an API and I'm done.
But the thing that I have thatmy competitor or another,
company doesn't have is I haveproprietary data.
(12:16):
I have enterprise data, so Iknow I have more information and
can actually tailor theseproducts to offer a better
customer experience.
I'll give you an example.
If you work with an airlinecompany for example, or if you
work with a hotel chain I canoffer a recommendation product.
For example, Hey maybe youwanna.
Have this for lunch or have thisexperience at the hotel, I can
(12:39):
offer a better solution or amore personalized concierge, if
you will, because I know yourpreferences and I have prior
information about what's yourlunch preferences or whatnot.
And so I can actually, I.
Make more intelligencerecommendations for you.
If that data is unreliable, ifI'm using the wrong data to feed
those AI products, thenobviously everything crumbles,
(13:01):
right?
So that's the second kind ofcore issue.
Make my data AI ready.
And a lot goes into that.
We can go into more detail.
There's you have to make surethat you have structured data is
reliable, your unstructured datais reliable.
That this whole thing is like abig thing, right?
But, that's the, second coreissue.
(13:21):
Then the third core issue that Ireally see data and AI leaders,
struggle with or find the needto divert attention and
resources to is now that they'vereleased AI products, how do we
make sure that those arereliable?
So how do we make sure that theagent is not selling the Chevy
Tahoe car for$1?
(13:41):
Or because you can have theperfect trainee data the perfect
prompt, the perfect context, butthe output of the model will
still not be fit for pur forpurpose.
And again.
Can go into more detail and sothat, but at a very high level
when I think about decisionmaking for a company and for,
budgets, these three things arehonestly table stakes.
(14:04):
I would say the first one,certainly like making data teams
more productive.
Yes.
With AI and automation.
Really table stakes.
I think the second more and moretable stakes, like getting the
foundation ready, making, havingAI ready data.
I think the third category,building reliable AI solutions
is probably an area that peopleare still in the early days of.
We're still many companies arejust moving to the cloud, right?
(14:29):
They're still in that, on thatjourney and there's a lot that
goes into actually delivering AIsolutions.
But at a high level, these arelike the three, three big
problems that the data and AIleaders have to tackle.
Andreas Welsch (14:41):
I think that makes it very tangible, right?
And also shows this progressionof how, should you think about
this?
Also in, in three simple stepsnow.
You mentioned yes we, need toput more emphasis on, data.
We need to empower data teamsmore.
What are you seeing, how areleaders getting buy-in along the
chain of command to do thesedata projects?
(15:01):
Again, when everybody's askingwhat's our AI strategy?
Barr Moses (15:04):
Yeah, it's a good question.
Look I think one of the thingsthat is really important and
obvious, but like easier saidthan done is tying what this is
to.
Tying what you're building to,to real value and to real impact
and maybe even just saving, cofinding cost savings along the
way using AI.
(15:25):
And by that lemme just give youlike a practical example of how
you can use AI to save time.
At Monte Carlo, what I mentionedwe offer observability for
companies or help companies makesure that their data and AI
products or their data and AIestate is reliable.
And we also use AI ourselves.
(15:45):
And so we actually built or weare building a number of agents.
We're building an observa suiteof observability agents.
And the goal of this agent isbasically to deliver real hard
ROI for data and AI teams.
And, I'll give you an example orkind of walk you through how
these works both'cause I thinkit's super cool.
(16:06):
But also because I think, Iactually think it's valuable for
data and AI team.
One of the things that datateams spend a lot of time on,
mostly data analysts spend a tonof time on, is trying to figure
out what in their data needs tobe monitored.
So I have a lot of tablessometimes thousands, tens of
thousands, hundreds of thousandsof tables in my.
(16:27):
Data lakehouse into my datawarehouse.
That could be at your GCPplatform, Azure, Databricks,
Snowflake, AWS, what have you.
And if I'm a data analyst or adata engineer, I need to make
sure that data is accurate, butI have no idea.
I.
What are their requirements?
It's hard for me to be able tospecify at the table or even
(16:48):
feel level to know exactly whatneeds to be accurate, what the,
definition of accuracy evenmeans.
And so oftentimes what datateams do is they spend time
rigorously and tediously, firstof all, profiling the data.
So like understanding thestructure of the data and
understanding the data itself.
Digging through that, lookingthrough metadata, making
(17:10):
connections to understand thesemantic meaning of fields.
And then coming up with specificmonitors that might say, oh, you
know this, I'm just making thisup.
Again.
If this is like a, an airlinethe, and I have a column that
has number of every row is aflight.
(17:30):
And so I need to make sure thatthe flight numbers have a
certain sort of they need tolook a certain way, they need to
have a certain number ofcharacters, et cetera.
So I need to have a monitor tomake sure that that data is
always accurate.
'cause I can't mess up flightnumbers, right?
Yes.
Like flight numbers, data needto be accurate.
And so that process is really,hard and tedious and manual to
(17:54):
go through all of that and comeup with monitors.
What we have done is actuallyreleased a monitoring agent.
It's in production already.
It's used by hundreds ofcustomers, so it's been out live
and for, a while.
We actually have a 60%acceptance rate, which means
that 60% of the monitors that werecommend are being used and get
accepted.
And what we do is actually, thisagent actually weird to say
(18:18):
this, but it mimics the humanbehavior.
It goes through, it profiles thedata, it tries to make this the
understand the meaning ofdifferent connections between
the data and then makerecommendations for what you
need to monitor.
And so that cuts down for a datateam time from like weeks to
(18:39):
minutes if you will.
And so it's really cool to startseeing what you can actually
start offering to data and AIteams to make them more
productive.
And so what we're building andwe're about to release in the
next few weeks.
And this is what I think is areally cool application of LLMs
has to do with troubleshooting.
So the first thing that data andAI teams do is they monitor
(19:01):
data, right?
They need to know when thedata's wrong.
The second thing that they do iswhen the data is wrong, they
need to start.
They have a workflow forunderstanding why it's wrong,
starting to triage andtroubleshoot and try to
understand what was the rootcause.
And so what our, troubleshootingagent does is again, mimics what
a data gov data steward woulddo.
(19:22):
What a data steward data analystwould typically do is they would
start with coming up withhypothesis.
For what might be wrong.
So let's say I get anotification that there's a
particular issue with aparticular report.
And this report is being usedevery single day by our field
operations team.
It's high visibility.
(19:42):
It's really important that I.
Look into this issue.
It's, data.
It's a report that's being usedall the time, and then I start
coming up with hypothesis forwhat could go wrong.
And I start going upstream tableby table and saying, okay, let
me check if the data arrived ontime.
Okay, let me check if thisupstream data arrived on time.
Okay, now let me check thatanyone make a change to the
(20:03):
code.
Somewhere that anyone breaksomething.
Okay.
Maybe the ETL system, maybe thethe job failed and then
incomplete.
I basically had to come up witha list of dozens of hypothesis
and start to check them and testand cover what happened.
What we've done in, thistroubleshooting agent is we
basically have an ensemble ofLLMs.
(20:24):
There's an l there's this masterLLM that comes up with the list
of hypothesis, and then itspawns off a new agent for every
single hypothesis.
And so every agent thenbasically recursively looks into
a particular hypothesis, and soyou can have up to, I think
around a hundred or so agentsrunning in less than two
(20:45):
minutes.
All looking into differenthypothesis for what could go
wrong at the same time.
And so something that could havetaken me years frankly, to go
through all of this hypothesisand try to understand what goes
wrong with a troubleshootingagent, again, with a combination
of.
Breaking down the problem intodifferent tasks.
Every task is a hypothesis tolook into.
(21:08):
And then using a lot of LLMs tolook into each of this
hypothesis and then basicallycome back and report back on
what they found.
And then this sort of master,there's a, there's sort of a.
Bigger LLM that kind ofsummarizes all that and gives
you like A-T-L-D-R, here's theroot cause what happened this is
the reason for this incident isX, Y, Z.
(21:29):
And so that is really powerful.
So you're basically taking,again, things that data and AI
teams are already doing, likethey're already spending time on
that basically cutting theamount of time significantly.
And so the ROI in thoseinstances is very clear, if that
makes sense.
Andreas Welsch (21:46):
Great example.
And very tangible too.
I love these examples where it'sabout, hey, it would have taken
one person or a team of peoplemonths or years to do this.
Now we can do this in minutes oreven less than a couple minutes.
So super powerful, and I thinkwe need more of these examples
that show what can this actuallydo.
(22:07):
Beyond the hype talk and totallythese kind of things.
So sounds like you are alreadyworking on some exciting things
and are testing this withcustomers or rolling this out to
customers.
Super, super cool.
Now, it's been a few weeks.
I think I would say it'sprobably even been a few months
(22:27):
since we've started talking andsetting up today's session.
And I remember probably earlierin the year I came across an
article from you where it saidAgentic AI might actually be
doomed to fail.
And we, somehow landed on thistopic of getting data AI ready.
But I'm wondering why do youthink AI agents might be doomed
(22:48):
to fail?
When are they doomed to fail?
It's probably the betterquestion to ask.
When are they doomed
Barr Moses (22:53):
to fail?
Yeah, good question.
Look we talked about thisearlier on this call when folks
are releasing hundreds of agentsinto the wild and like waiting
to see what happens.
I think we know what happens.
I think we know what the storylooks like here.
And I love this example of sortof the troubleshooting agent
(23:14):
because I just wanna stay, Ijust wanna say when obviously
when our team presented to me,shared we were working on the
troubleshooting agent.
I think that was like a turningpoint for me because I think,
there's a lot of AI skepticismout there and there's a lot of
question of what does itactually mean?
(23:34):
And just to be clear I don'tthink AI will replace people or
replace data and AI teams.
If anything if I look at acorollary like engineering, as
engineering as a space, the moreadvancements we've had with
engineers the higher there was ademand for engineering teams.
And I think the same will be fordata and AI teams.
And the more we use AI, the morewe need data and data, there's
(23:57):
gonna be an increase in demand.
It doesn't mean that we don'tneed to change how we work and,
what we do.
And I'm, I very much believe inthat.
And also I think thisapplication of the
troubleshooting agent reallybrought to life for me What.
How powerful if you put LLMs touse, how powerful that could be.
(24:20):
And so you can imagine a lot ofcomplex problems that you might
be working on today that if wecan intelligently break them
into smaller tasks and usedifferent, smaller and larger
LLMs in different instances, youcould actually teach LLMs to do
really, clever things.
And so I wanna start I, thinkit's important me to say that
because I think.
(24:41):
I think figuring out what agentscan do for your business is
important.
And I do believe, I am convincedtoday that there is value in
that for your company, for yourbusiness.
And I think you have to see itand to experience it yourself,
to believe it.
It's very different when it's intheory.
(25:03):
So first of all I and, there's alot of talk about like, how in,
in a couple years, softwareengineers will really just be
managing a fleet of agents.
Like you'll have one agent tobuild features, one agent to fix
your bugs, one agent to do PRDswith you basically have a fleet
of agents.
And so similarly you canenvision that in the world, data
(25:24):
and AI teams have like a fleetof agents one agent to monitor
their data, one agent totroubleshoot their, and we'll
all just sit here in our cushychair and monitor all the
agents.
I can't wait for that.
I obviously like that means thatour world will, be very
different and how we work willbe very different, but, in that
(25:45):
world, or the reason why I wrotethat, that I think there's a
high chance that AgTechsolutions will fail is if we,
are not thoughtful about what itmeans to build high trust
agents.
And I'll give you an example,the the most like popular
(26:09):
example, if you will think thiswent viral on XA couple years
ago.
Or, maybe a couple last year, Ithink someone wrote on, on one
of sort of Google's LLMsolutions.
What should I do if cheese isslipping off my pizza?
Yes, you may have seen this andthe answer was, oh, you should
(26:30):
just use organic super glue.
No problem.
To like glue the cheese backonto your pizza.
Now, just to be clear, like thetraining data was good.
The like prompt was good.
The con, everything was good,right?
I.
But the output was totallyinappropriate and made no sense
for the context.
Now I will continue to useGoogle or maybe use perplexity
or something like, but Googlecan get away with it.
(26:53):
We can't get away with that.
Most organizations in the worldcan't get away with building
agents that will have, that willshare responses like that.
That are really inappropriate.
And, continue to have trust.
And so I think the way that weneed to manage our agents and
(27:14):
the way that we need to manageour data and AI platform needs
to change drastically.
And one of the things that we'vedone at Monte Carlo over the
last several years is work withthousands of enterprise and help
figure out.
What is the way to buildreliable data and AI products
(27:34):
and based on, again, thousandsof, customers and conversations,
we've found that problems withdata and AI products can really
be reduced to four coreproblems.
The first is data and AIsolutions can be running on bad
data, meaning the data thatyou're feeding, it might be
inaccurate or late or justwrong.
(27:56):
The second core reason for whydata and AI systems might be
wrong is because there's a codechange.
Maybe it's a bad join maybe it'sa schema change.
Maybe it's a code change in theagent itself that's actually
changed something dramatically.
So the third reason for whythings can go wrong is if
(28:18):
there's a problem with thesystem.
A system that could be like yourETL jobs, like airflow or DBT,
or it could be orchestrators, itcould be things like land chain
or land graph.
Basically anything along yourdata and AI say can go wrong in
your system.
'cause the one thing for sure, ahundred percent of systems fail
at some point.
(28:39):
And then the fourth thing thatcan go wrong, and I alluded to
this earlier, is.
You can have all those thingswork perfectly.
You can have the perfectcontext, the perfect prompt, but
still the map model output willbe not fit for purpose.
And so I think for people whoare building agents who are
building AI solutions, unless westart thinking about the
(29:00):
holistic health of those systemsoverall as a function of those
four things, data code systemsand model output, we will fail.
Andreas Welsch (29:11):
Wow, that's that really hits home and I think
comes down to the core of itreally.
Barr, thank you so much forsharing this.
We've covered a lot of ground inthe last 30 minutes.
From helping data teams empowerthem and understand that they
have big responsibility in whatthey can do to observability,
(29:32):
making sure that our data isactually accurate and correct.
And then talk about some of therisks, what happens when you
don't do that, and where thefailure points in, Agentic AI.
Also, great to hear what you'redoing at Monte Carlo and how
you're bringing AI and agents todata and AI teams.
Sounds like a really excitingopportunity and really exciting
space too.
Barr Moses (29:50):
So yeah.
Thanks.
Andreas Welsch (29:53):
Yeah, wonderful.
Thank you so much for joining usand for sharing your experience
with us today.
Barr Moses (29:57):
Absolutely.
Thank you for having me.
It's an exciting time.
There's a lot of exciting stuffhappening, I think the future is
bright.
I'm excited to be part of it.
And thank you to everyonejoining.
Popular Podcasts
United States of Kennedy
United States of Kennedy is a podcast about our cultural fascination with the Kennedy dynasty. Every week, hosts Lyra Smith and George Civeris go into one aspect of the Kennedy story.
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
Bookmarked by Reese's Book Club
Welcome to Bookmarked by Reese’s Book Club — the podcast where great stories, bold women, and irresistible conversations collide! Hosted by award-winning journalist Danielle Robay, each week new episodes balance thoughtful literary insight with the fervor of buzzy book trends, pop culture and more. Bookmarked brings together celebrities, tastemakers, influencers and authors from Reese's Book Club and beyond to share stories that transcend the page. Pull up a chair. You’re not just listening — you’re part of the conversation.