July 22, 2025 • 26 mins
In this episode of The Disruptor Podcast, host John Kundtz sits down with Harish Chandramowli, founder of Flaire Software, to explore how AI is revolutionizing traditional ERP systems.
Drawing on his background in cloud security at MongoDB and Bloomberg, Harish shares how his deep technical expertise and hands-on experience led him to create Flaire, a lightweight, AI-enabled workflow optimization platform specifically designed for the fashion industry.
Harish walks us through how most SMBs are stuck trying to force their business workflows into rigid ERP systems built for enterprise-scale complexity.
Instead, he advocates for a radically different approach: utilizing AI and platforms like Shopify to create customized operational workflows, without the overhead and cost associated with traditional ERP deployments.
We discuss:
- The evolution from cloud security to fashion-tech entrepreneurship
- What a “schemaless ERP” really means—and why it matters
- The hidden costs and traps of customizing legacy ERP systems
- Real-world examples of AI-driven workflow automation, including SKU creation and product launches
- Why knowing your internal workflows is critical before automating with AI
- How AI is changing the speed, cost, and expectations around operational software
Harish’s philosophy is clear: Don’t overbuild. Instead, understand your business deeply, build simple, iterate quickly, and only automate what’s ready for it.
Whether you're a founder, operations leader, or curious technologist, this episode offers a fresh take on how to streamline your back office with modern tools and AI.
To learn more about Harish and Flaire Software, visit their website (flairesoftware.com) and connect with Harish on LinkedIn.
Comments or Questions? Send us a text
***
Engage, Share, and Connect!
Spread the Word: Valuable insights are best when shared. Share this episode with peers who may benefit from it if you find it insightful.
Your Feedback Matters: How did this episode resonate with you? Share your thoughts, insights, or questions. Your engagement enriches our community.
Stay Updated: Don’t miss out on further insights.
Subscribe: You can listen to our podcast, read our blog posts on Medium, Substack, and LinkedIn, and watch our YouTube channel.
Collaborate with The Disruptor and connect with John Kundtz.
Got a disruptive story to share? We’re scouting for remarkable podcast guests. Nominate a Disruptor.
Thank you for being an integral part of our journey.
Together, let’s redefine the status quo!
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
John Kundtz (00:00):
Disrupting traditional ERP implementations
with AI-driven workflowinnovation.
Hi everyone, I'm your host,john Kuntz, and welcome to this
(00:21):
edition of the Disruptor podcast.
For those that are new to ourshow, the Disruptor Podcast.
For those that are new to ourshow, the Disruptor Series is
your blueprint forgroundbreaking innovation.
We started this podcast inDecember of 2022 as a periodic
segment of the Apex Podcast.
Our vision was to go beyondconventional wisdom by
confronting the status quo,exposing the raw power of
(00:43):
disruptive thinking.
Confronting the status quoexposing the raw power of
disruptive thinking.
Today, we will be talking withthe founder of Flare Software as
we explore how organizationsare optimizing their operations
without investing in traditionalERP systems and how AI is
making all that possible.
Welcome to the show, harish.
Glad to have you on.
Thanks for having me.
(01:04):
Welcome to the show, harish.
Glad to have you on.
Thanks for having me.
Tell us a little bit about yourbackground, from cloud security
to Bloomberg and MongoDB, andwhat led you to start Flare
Software.
Harish Chandramowli (01:13):
I started at Johns Hopkins for cloud
security.
It was really amazing, right.
But I also noticed that a lotof my friends went on to go and
start, like some of the amazingcompanies out there.
At the time, entrepreneurjourney was not suitable for me
due to personal reasons, butthat's when I joined Bloomberg.
(01:34):
One of the things I came whilecoming out of cloud security at
Hopkins for me that stood out isthat, like I never wanted to be
a consultant, I always wantedto build and I wanted to be a
developer, so I joined Bloomberg.
When I joined Bloomberg, theysaid like, hey, there is this
niche field where you buildauthentication system, you build
(01:54):
the encryption systems, so youneed the security knowledge, but
you are a software engineer.
So that's how the journey atBloomberg started From there.
Mongodb Atlas just came out andthey were trying to hire someone
for security so that they canstart building the security team
, and that led to MongoDB.
Again, the team was pretty muchthe same, like building
(02:16):
products over, like saying thatI'm an expert in something, I
guess.
So I joined them as a securityengineer.
I moved around different teamsbased on like what security?
What they needed some securityexpertise.
That means I got the breadth ofknowledge.
I got to build products fordifferent teams, which led me to
be joining as a lead engineerfor one of their core Atlas
(02:39):
cloud team and during those fewyears as a lead as on-call
person, I also observed how datais being used by my customers
and some of our biggestcustomers are in retail.
So that made me more and morecurious on how data is being
used in the retail industry andwhy that's a big thing in retail
(02:59):
industry.
And obviously when I looked atNetSuite, oracle being like the
biggest player in like an ERPindustry made me even more
curious.
It's a database company.
Why are they shedding so muchmoney in this industry?
That's kind of like mycuriosity around the industry.
The way it started was that Iwas in this shop called ONS in
(03:21):
Soho, new York downtown.
I was observing how they wereoperating.
I was talking to their founders, I sat in some of their team
meetings and the more I observedit felt like the back office of
fashion is a workflow and dataproblem.
When we think about fashion,fashion has a lot of technical
softwares involved in it.
Starting, which is a littlemore obvious to everyone, is
(03:42):
design, because you kind of tryto figure out what is the
embroidery, how bunch ofdifferent embroidery styles can
get put together and visuallyhow it looks before production.
The moment that happens.
Then the process of procurement, ordering, negotiating
different factories, figuringout how to order, how much cash
you are committing to differentproducts All those complexities
(04:03):
come into place and that's whereERP lifecycle begins.
You can think of ERP assomething that brings various
moving parts inside the companytogether.
So the first part I talkedabout how ERP comes brings the
data from the design team intoproduction lifecycle.
The ERP again, once a productis received is the one that
(04:23):
pushes and helps maintain yourinventories and then, like, put
listed in different websitesshopify, amazon or whatever it
is and final piece is like youmake an order in your website.
Then erp again comes into playon, like making sure your order
goes to warehouse and warehouseships it properly and then you
get back the tracking number.
So erp kind of plays a verycrucial role in your whole
(04:46):
workflow and how data moves,even to place a single order in
your in someone's website.
And that made me more curiousand, having coming from a data
background, having built a lotof data products, I felt like
this area is ripe for innovationand some of the questions you
usually ask, like I was happy atmongodb, I was happy at
bloombergs.
(05:06):
So the first question you alwaysask if I want to do this, why I
want to do this.
My question always was likeshopify has done a tremendous
job of e-commerce, especiallywhen it comes to sales platform,
and I kind of don't want to dosomething over there because
when someone has done a greatjob, I know I cannot as a single
person, as a startup, do abetter job than shopify.
(05:28):
Oracle has done a great job forbigger industries 100 plus
million band who needs, like alot of engineers, customizations
.
Yes, I am not going to disruptthe engine, that industry, at
least on day one.
But then I also realize thatthere is like smb market who
can't spend that much onsoftware engineering but can a
lot out of data by bringing ininnovation in this field, and
(05:50):
that's what led me to startFlare.
John Kundtz (05:52):
One of the things I love about this podcast is I
interview people that have thesebackstories, that sort of go
weaving around, and eventuallythey get to where they are today
.
That's fascinating.
Eventually they get to wherethey are today.
That's fascinating.
And it's cool that you keptbuilding upon your knowledge and
(06:12):
your desire to figure out howthings work on the back.
So describe to me Blair yousort of went into this a little
bit but describe to me the wayyou use your software to avoid
this.
Traditional ERP systems because, as we know, as you mentioned,
(06:34):
I mean ERPs for largeenterprises is hugely expensive.
It's uber, time consuming andit's also, once you're in,
you're sort of locked in right.
So if you go down the Oraclepath, the switching costs are
huge once you implement it,Certainly from my experience in
the larger enterprises.
Talk a little bit about whatkind of mistakes you see
organizations making when theytry to scale with traditional
ERP platform.
Harish Chandramowli (06:55):
So I'm going to kind of spread it out.
At the beginning I'm going totalk about how Flan came up in
terms of innovation and make iteasy.
But obviously over the twoyears with it and everything
what we are building as well,like how I see the industry
keeps evolving.
That's not the important thing,right?
You need to keep up with theindustry and bring in that
innovation.
Let me start with when Istarted and why it was very
(07:18):
useful for my customers to comeinto us and even do that
migration.
You are right, doing an erPmigration is a big, big step and
trusting a startup to do islike even bigger step.
Our customers like Alala, rta3,all of them are like 30-50
million brands, not like a newshop that came up.
So this is how it started,right?
(07:38):
One of the thing I noticed whenI was in back office is that
people see each of your dressand want to analyze whether to
buy it, whether to replenish itin a different.
If it's a t-shirt, it's justsize and color, so you want to
see which color is moving fast,which size I need to order more,
which size I need to order morefor me, or so that makes sense,
whereas when you come to femaleathletic wear, then it becomes
(08:01):
more complicated.
You have cup size, you havetorso length, you haveor and the
regular sizes.
So now you are looking evenmore matrices to make your plan
to do analysis around what isselling fast, what is not
selling fast for a beginningseason, what I have to do.
So that's why the attributioncame into play.
That was one of the biggestreason people are moving In
(08:23):
traditional ERPs, even in likesee, if you want these custom
attribution, you need to pay fora software engineer, you need
to customize it, and when Ilooked at it, for me just felt
like an extension to MongoDB.
In typical database you think ofprimitives as numbers and your
alphanumeric parameters, butwhen I speak more to these
(08:45):
vertical, specific industriescolor, for example it's like a
primitive data type to them,barcodes, which you see at the
top it's like a primitive datatypes.
So what I ended up doing islike extending MongoDB's
functionalities to include theseprimitive types which each of
the brands, and from that webuilt a schema-less ERP.
That means we don't chargepeople for customization.
(09:07):
You come to us and say, hey,this is exactly how I look at my
business, these are theproperties I want to be tracking
for my clothes and we set themup.
That was compelling enough forthem.
They were willing to take therisk to do the data issues.
That's pretty much the reasonwhy people started and kind of
like it's important that whenyou are you, when you are in
(09:29):
your business, you get to makedecisions that makes your
business better and not changeyourself for the software that's
available, software that'scheaper.
That's first one.
Second one we did.
I started it as fully apparelspecific and even if you are not
in Shopify, we were building it.
But the more I started buildingit, the more I noticed is that,
(09:51):
unlike traditional ERP, wecould tightly couple with
Shopify.
Shopify has done a tremendousjob of exposing APIs and
whatever Shopify does well, Idon't want to repeat.
I just want to pass on thatload to Shopify.
What this meant at the end ofthe day for my customers is that
they end up being somethinglike a Shopify app.
Once you migrate to us foranother migration or if you want
(10:15):
to use another new software,you can use any of the software
in Shopify app ecosystem.
You don't want to.
Erp never becomes a bottleneckfor you to try different apps
marketing apps, offer apps,bundles, apps whereas with
traditional erp.
When you want a new app,because erp is not tightly
coupled with shopify, then yourerp needs to support those
(10:37):
integrations.
So that means you are notnimble, you are not fast and,
let's be honest, not every erpcan build an ecosystem that shop
has well, the amount of adsthat shoppers and that's the
other part of how I see that wecould build something that's
different while giving users thepower and bigger
functionalities.
John Kundtz (10:57):
So, in a nutshell, I think what you're trying to
say is you're sort of bringingtogether the I call it the back
office and the front officeright.
So the back office is all theERP, traditional inventory
management and orders and allthat stuff that goes in the back
end.
The buyer right never reallysees that.
They see the front end, whichis the Shopify, and so you're
(11:17):
bringing both of those together.
I think and that's probably themistake I'm hearing that maybe
organizations might make.
Harish Chandramowli (11:24):
Absolutely right.
Traditional ERPs used toreplicate what our data is in
shop.
If I ask them in the backoffice, that is disconnects and
everything we try to be like.
Let's bring things togetherbecause it's a single stack.
Why are you thinking as eachand everything is very different
?
John Kundtz (11:38):
Which sort of leads us into the next question we
were talking about in the prepmeeting was this whole idea of
workflow optimization a betterpath forward?
So what does that really meanin practice for, let's say, a
small or mid-sized organizationthat you might be working with?
Harish Chandramowli (11:55):
Yes.
So in terms of workflowoptimizations, a lot of people
have very custom workflows.
Some of them print labels Forsome of them, baha House prints
labels and each and every onelook, even the headcounts they
have.
The type of people they have asbusiness evolves is very
different, especially the biggerthe business, once you cross
that 10, 15 million mark.
(12:16):
Now you have an operationperson and you have a production
person who is different from anoperation person.
Now you have a customer serviceperson who is different from
these people.
How do you bring everyonetogether?
One of the things that a hasdone a really great job is being
able to build theseintegrations based on various
tooling.
You use various workflow.
You have the customizations atthe top, the last.
(12:39):
At the beginning, I usuallytell people to change their
workflow so that you adapt tothe software and be more
efficient.
But in the last six months toeight months especially when
people come and say, hey, thisis my workflow, I need this
customization I was able to goback and build it in two days,
just like the smaller changesand workflow optimizations, and
that's how I also see thingsevolving, moving forward and
(13:06):
being able to build these alittle more in a nimble, agile
and quicker way that you canactually go to people and say
this is my workflow, this is mybusiness context and engineers
come up with the software thatadopts to your workflow and your
business context and you don'thave to pay $100,000 to $100,000
for implementation.
John Kundtz (13:24):
That makes sense.
In the old days, you would tryto avoid customizations, correct
, because every time you did anupgrade, every time you made a
change, the custom integrationswere always what helped people
from moving from one version ofenterprise software to another.
It's not here that AI isallowing you to do that quickly.
Therefore, you can bring theboth worlds together.
(13:46):
You could quickly deploycustomizations that aren't going
to necessarily bite you in thebackside in the future when you
have to make upgrades or changes.
Is that sort of what you'resaying?
Yes, absolutely right.
So walk us through a real lifeexample of how that works.
Harish Chandramowli (14:04):
For me, the real life example is SKU number
generation.
Everyone generates SKU numberwith some kind of custom logic.
People were using spreadsheetto generate those SKU numbers
and I was able to just write aquick custom logic in the front
end that says hey, for thiscustomer, this is how they
create SKU numbers.
Like, if you have color red,that's number 48.
(14:25):
If you have that, if the cupsize is x, c, add c to n.
All those customizations now isso much easier that people don't
need to manually do it and youget your own specific things.
Number one.
Number two for example, if youwant people use project
management tools to createdesigns and everything, and the
moment everything gets approved,they go to shopify and then
(14:46):
they create.
The moment everything getsapproved, they go to Shopify and
then they create the product inShopify.
You don't need to do that AA.
You can build thatcustomization Once in my project
management tool.
I am approved directly, just goand create the product in
Shopify.
So those kind of biggercustomizations.
It used to take a week becauseyou need to know the APIs for
(15:07):
Shopify, you need to know theAPAs for these project
management tools.
Aa makes it easy.
Number two we always thinkabout product, as all the edge
cases, we need to fix somethings.
In fashion industry or in mostof the vertical specific
industry, people do the exactsame workflow.
You just need to make suretheir day-to-day workflow works
(15:29):
for these custom software andthat, so that means testing is
easier.
Users become your testers.
They're happy and you get allthese kinds of workflow
automations in a matter of days.
John Kundtz (15:41):
It leads me to the following question In your
experience, your clients or yourcustomers, do they understand
their workflow in the beginning,or is that part of the upfront
process I've been reading andlistening to and observing and
even experiencing is?
Ai is super powerful, but youhave to know what you're doing.
You have to really take thetime and understand what your
(16:04):
workflow is.
I mean, you can't automatesomething if you don't know what
it is.
Is that making sense?
Is that right?
Harish Chandramowli (16:11):
It's absolutely right.
It's one of the places that Iuse AI to meet myself more
better at sales.
So one of the things I do islike exercises I do in the first
call is ask them explain to mepurely in business terms, don't
explain to me how people useyour ERP.
I don't want to hear that frombusiness terms.
Explain me what people you haveand what is the functionality
(16:33):
for each one, and then I feed itto Android, which gives me a
really smart my JS workflowchart.
Then I sit with my customersand be like this is a very
visual workflow chart.
This is what I understood fromyou.
Can you validate?
This is how your businessoperates.
Then it becomes more easy tosay how my software fit into
their workflows.
John Kundtz (16:54):
So I love that for actually two reasons.
One is what you're describingto me is something I would call
you're co-creating with yourclient or prospect, so you're
trying to understand theirbusiness, probably better than
they may understand it I knowwhen I've done this.
A lot of times I've laid stuffout like that and the client
(17:15):
looks at me and goes, wow, Inever knew that before.
I don't even know that aboutourselves and I work here every
day and now it also allows youto follow up with them correct
and start to build thatrelationship.
One thing hasn't changed in thelast 40 years that I've been
selling is people still buy frompeople they trust, especially
(17:36):
if you're new, you're anentrepreneur, you've got a
startup, whatever, but itdoesn't matter.
You can be a large or smallcompany.
At the end of the day, they gotto look at you in the eye and
say, hey, Harish, I believe youcan help me, and you don't do
that just by sending them abunch of emails or just assuming
(17:57):
you know what they're doing.
I think that's great.
Let's circle back on the AIplay.
Ai is effectively being usedwhat you do without
over-promising what you can do.
I mean you described a littlebit about it in the pre-sales,
but over on the back end orthrough the course of the
buyer's journey and ultimately,the implementation of your
software buyer's journey and,ultimately, the implementation
(18:19):
of your software.
Harish Chandramowli (18:24):
So one of the things that was very
important, even as a CTO andfounder, is to keep an eye out
of which evolution of AI isproduction ready versus which is
not.
We hear a lot about agents, howagents can do everything for
you in production, but think ofsomeone getting an invoice,
keying the invoice in to makethe payment.
If an agent is doing it, if itmakes a mistake, then the
repercussions are big.
You are not paying the rightamount, your books doesn't tally
(18:46):
, then you need to go throughlike thousands of invoices to do
it.
But so agents is not there forthat specific use case, whereas
the use case where, example,some people in our industry miss
emails that says purchase orderis getting delayed, I could
easily parse every email thatcomes in.
When I feel something seems tobe off for their manufacturing
(19:09):
communication, I can flag thatemail at the top so that people
can go and verify it.
And a is that a can do thatreally well.
So understanding which part ishype and it and AI is that AI
can do that really well.
So understanding which part ishigh and which is production
ready is really important andwhich is very critical to users
is important.
The second part isrepeatability.
Right, lot of times when I talkabout invoice passing or
(19:30):
uploading your customer ordersto understand how they operate,
some of them needs to berepeatable and you need to be
sure that when you repeat it itwill give you the exact same
results.
And AI doesn't do that.
But what AI can do is if I useAI to build those customizations
as an app, if I do it to buildall the custom logics, then you
(19:51):
can test it, deploy those customlogics and productions.
You know that when you areusing the product it's not AIs
they'll go and give you theexact same results, whether it's
a SKU number generation, like Imentioned, or creating some
custom PDF to send to yourcustomers for invoicing All
those things.
Once you use AI to build itfaster, you can use it in
(20:12):
production Agents.
Yes, email passing agents exist.
If you come up with a invoiceor, like we call it thing, list
in our industry which comesmanufacturers and what item they
are producing and how much, itused to be hundreds of lines.
Someone manually used to enterthose key in those things in erp
, but now you can upload it.
(20:32):
We pre-pass but we don't reallyenter the value.
Rather we ask someone to verifyit.
Now your job of keying in forone hour changes to a job of
five minutes of verifyingwhether the past values are
right.
So that's how I see ai, andobviously a month from now, like
you mentioned, if you ask me,maybe agents is there, I would
say agent would do the wholething.
(20:53):
It's all the hype.
You can sell a dream, but youalso need to be understanding
which is production ready, whichcan be used in real life today,
and not just what is going tocome.
John Kundtz (21:03):
Yeah, I'm glad you hit on agents, because it drives
me crazy, because if you watchthe external marketing and
newscasts or whatever,everybody's talking about agents
and, as you mentioned, it'scoming, but it's not the panacea
, especially today.
So great advice.
So, speaking of advice, tell us, as a founder, if there are any
founders listening today, whatadvice would you give to them
(21:25):
about building the rightoperational stack, particularly
when it comes to technical?
We're all overwhelmed with techoptions these days.
Harish Chandramowli (21:33):
Keep it simple.
Pick things that you are reallycomfortable with.
I even while working atmongolian stuff, while I have
seen people how they usedatabases, my first thought is
that unless you are like thecoinbase and google of the world
, the reality is it doesn'tmatter, even if your query is 10
milliseconds slower, your userswon't do it.
So pick something that you arecomfortable.
(21:55):
I mean obviously.
Obviously there's other changesif you're building a product,
but when you are building theseinnovations in ERP or these kind
of like very real-worldproblems, trying out different
ways, trying out differentthings helps.
People needs a lot ofvibrations.
So pick something that's soeasy to you that you can trade
(22:16):
until you see a market fit.
Then, obviously, you can justgo back and upgrade everything.
John Kundtz (22:21):
That's great advice , I agree.
I see people, especially in thetechnology space, just trying
to create solutions purely onthe technology, and what I
really have picked up throughthe course of this conversation
is you're really focusing onwhich is what I advise all my
salespeople to do and all thetechnical people that I advise
(22:42):
the product managers of theworld and stuff like that is
really focus on the why.
Why is what you're buildingimportant to, ultimately, your
end user or your buyer or yourcustomer or your client?
And then build it quickly andsimply and then iterate over and
over and over again.
Harish Chandramowli (23:04):
Yes, One example I would usually give
people is like when I talk aboutemail passing, I can make sure
Google can tell me when a newemail arrives, or I can just
query every 10 minutes onwhether this person has new
emails.
The later one is quick to build, not the most efficient way,
but until you prove out thatit's useful, just in the
quickest way.
John Kundtz (23:24):
That's great.
Yeah, it's a classic agilemethodology or mantras I like to
say fail fast and cheaply.
Right.
So build something.
Test it out.
If it doesn't work, throw itout.
Pivot If it works.
Great.
If it needs to be better, testit out.
If it doesn't work, throw itout.
Pivot If it works.
Great.
If it needs to be better, makeit better.
All right, before we wrap up,switch gear.
What's something you've learnedthe hard way in your experience
(23:45):
that has shaped you to run yourbusiness today?
Harish Chandramowli (23:49):
Personally I feel when I joined Hopkins I
had to take two jobs and likethe five grad courses that means
I kind of got so used toworking long hours and working
long hours is very, veryimportant as a founder at the
beginning stages because youneed to find the product market
(24:12):
fit quickly.
So that experience definitelyshaped that.
And over the last two yearsthings I learned as founder is
not saying yes to all thecustomers and being very
pinpointed in the type ofcustomers you want to onboard at
the beginning.
It's so exciting to get lost.
I got lost where some peopleknow who are not in shop.
(24:35):
If I have my, I love thesolution I want to onboard and
end up building a lot for them.
But looking back, no, I don'thave the resources.
I don't have the money beingleased and focused on a small
group, even though you can havea bigger vision, so important I
think that's great advice.
John Kundtz (24:50):
I've seen that mistake that you said you
avoided over and over again.
Where, especially as a newcompany or a founder is every
prospect is a good deal, ittakes a maturity level to say no
, we're not a good fit right.
And understand that you couldget just sucked down a rat hole
(25:11):
and blow all your cash on oneclient and therefore have so
many small early stage companies.
It's not that they don't havegood ideas.
It doesn't mean they have badmanagement.
It just a lot of times theyjust run out of cash and if you
don't have cash, unfortunatelyall the good things you do can't
happen.
All right, we'll wrap it up.
(25:33):
Want to thank you for sharingall these insights and
experiences.
How can people learn more aboutyou and your company and your
services?
Where do they go?
What are your socials?
What do you want to share withthem?
Harish Chandramowli (25:47):
I always tell people they can follow me
on LinkedIn.
I usually update, keep myLinkedIn updated.
You will see what we are doingin the company in my LinkedIn.
John Kundtz (25:55):
any thoughts I have on LinkedIn and if you have any
questions, you will see what weare doing in the company in my
LinkedIn, any thoughts I have onLinkedIn and if you have any
questions, you can obviously DMme and always happy to reply and
of course we will include thoselinks in the show notes so you
can check it out.
You want to learn more orconnect?
And, harish, I'll give you thelast word before we wrap up the
show.
Harish Chandramowli (26:15):
I'll give you the last word before we wrap
up the show.
One of the things I would sayis just have fun.
If you're picking a software,any job, whatever it is, just
say pretty much B-Bull, that'sawesome.
John Kundtz (26:26):
All right, bud.
Hey, I'm John Kuntz.
Thanks for joining us on thisedition of the Disruptor Podcast
.
Have a great day.
Thanks a lot.
Thank you.
Disrupting traditional ERP implementations
with AI-driven workflowinnovation.
Hi everyone, I'm your host,john Kuntz, and welcome to this
(00:21):
edition of the Disruptor podcast.
For those that are new to ourshow, the Disruptor Podcast.
For those that are new to ourshow, the Disruptor Series is
your blueprint forgroundbreaking innovation.
We started this podcast inDecember of 2022 as a periodic
segment of the Apex Podcast.
Our vision was to go beyondconventional wisdom by
confronting the status quo,exposing the raw power of
(00:43):
disruptive thinking.
Confronting the status quoexposing the raw power of
disruptive thinking.
Today, we will be talking withthe founder of Flare Software as
we explore how organizationsare optimizing their operations
without investing in traditionalERP systems and how AI is
making all that possible.
Welcome to the show, harish.
Glad to have you on.
Thanks for having me.
(01:04):
Welcome to the show, harish.
Glad to have you on.
Thanks for having me.
Tell us a little bit about yourbackground, from cloud security
to Bloomberg and MongoDB, andwhat led you to start Flare
Software.
Harish Chandramowli (01:13):
I started at Johns Hopkins for cloud
security.
It was really amazing, right.
But I also noticed that a lotof my friends went on to go and
start, like some of the amazingcompanies out there.
At the time, entrepreneurjourney was not suitable for me
due to personal reasons, butthat's when I joined Bloomberg.
(01:34):
One of the things I came whilecoming out of cloud security at
Hopkins for me that stood out isthat, like I never wanted to be
a consultant, I always wantedto build and I wanted to be a
developer, so I joined Bloomberg.
When I joined Bloomberg, theysaid like, hey, there is this
niche field where you buildauthentication system, you build
(01:54):
the encryption systems, so youneed the security knowledge, but
you are a software engineer.
So that's how the journey atBloomberg started From there.
Mongodb Atlas just came out andthey were trying to hire someone
for security so that they canstart building the security team
, and that led to MongoDB.
Again, the team was pretty muchthe same, like building
(02:16):
products over, like saying thatI'm an expert in something, I
guess.
So I joined them as a securityengineer.
I moved around different teamsbased on like what security?
What they needed some securityexpertise.
That means I got the breadth ofknowledge.
I got to build products fordifferent teams, which led me to
be joining as a lead engineerfor one of their core Atlas
(02:39):
cloud team and during those fewyears as a lead as on-call
person, I also observed how datais being used by my customers
and some of our biggestcustomers are in retail.
So that made me more and morecurious on how data is being
used in the retail industry andwhy that's a big thing in retail
(02:59):
industry.
And obviously when I looked atNetSuite, oracle being like the
biggest player in like an ERPindustry made me even more
curious.
It's a database company.
Why are they shedding so muchmoney in this industry?
That's kind of like mycuriosity around the industry.
The way it started was that Iwas in this shop called ONS in
(03:21):
Soho, new York downtown.
I was observing how they wereoperating.
I was talking to their founders, I sat in some of their team
meetings and the more I observedit felt like the back office of
fashion is a workflow and dataproblem.
When we think about fashion,fashion has a lot of technical
softwares involved in it.
Starting, which is a littlemore obvious to everyone, is
(03:42):
design, because you kind of tryto figure out what is the
embroidery, how bunch ofdifferent embroidery styles can
get put together and visuallyhow it looks before production.
The moment that happens.
Then the process of procurement, ordering, negotiating
different factories, figuringout how to order, how much cash
you are committing to differentproducts All those complexities
(04:03):
come into place and that's whereERP lifecycle begins.
You can think of ERP assomething that brings various
moving parts inside the companytogether.
So the first part I talkedabout how ERP comes brings the
data from the design team intoproduction lifecycle.
The ERP again, once a productis received is the one that
(04:23):
pushes and helps maintain yourinventories and then, like, put
listed in different websitesshopify, amazon or whatever it
is and final piece is like youmake an order in your website.
Then erp again comes into playon, like making sure your order
goes to warehouse and warehouseships it properly and then you
get back the tracking number.
So erp kind of plays a verycrucial role in your whole
(04:46):
workflow and how data moves,even to place a single order in
your in someone's website.
And that made me more curiousand, having coming from a data
background, having built a lotof data products, I felt like
this area is ripe for innovationand some of the questions you
usually ask, like I was happy atmongodb, I was happy at
bloombergs.
(05:06):
So the first question you alwaysask if I want to do this, why I
want to do this.
My question always was likeshopify has done a tremendous
job of e-commerce, especiallywhen it comes to sales platform,
and I kind of don't want to dosomething over there because
when someone has done a greatjob, I know I cannot as a single
person, as a startup, do abetter job than shopify.
(05:28):
Oracle has done a great job forbigger industries 100 plus
million band who needs, like alot of engineers, customizations
.
Yes, I am not going to disruptthe engine, that industry, at
least on day one.
But then I also realize thatthere is like smb market who
can't spend that much onsoftware engineering but can a
lot out of data by bringing ininnovation in this field, and
(05:50):
that's what led me to startFlare.
John Kundtz (05:52):
One of the things I love about this podcast is I
interview people that have thesebackstories, that sort of go
weaving around, and eventuallythey get to where they are today
.
That's fascinating.
Eventually they get to wherethey are today.
That's fascinating.
And it's cool that you keptbuilding upon your knowledge and
(06:12):
your desire to figure out howthings work on the back.
So describe to me Blair yousort of went into this a little
bit but describe to me the wayyou use your software to avoid
this.
Traditional ERP systems because, as we know, as you mentioned,
(06:34):
I mean ERPs for largeenterprises is hugely expensive.
It's uber, time consuming andit's also, once you're in,
you're sort of locked in right.
So if you go down the Oraclepath, the switching costs are
huge once you implement it,Certainly from my experience in
the larger enterprises.
Talk a little bit about whatkind of mistakes you see
organizations making when theytry to scale with traditional
ERP platform.
Harish Chandramowli (06:55):
So I'm going to kind of spread it out.
At the beginning I'm going totalk about how Flan came up in
terms of innovation and make iteasy.
But obviously over the twoyears with it and everything
what we are building as well,like how I see the industry
keeps evolving.
That's not the important thing,right?
You need to keep up with theindustry and bring in that
innovation.
Let me start with when Istarted and why it was very
(07:18):
useful for my customers to comeinto us and even do that
migration.
You are right, doing an erPmigration is a big, big step and
trusting a startup to do islike even bigger step.
Our customers like Alala, rta3,all of them are like 30-50
million brands, not like a newshop that came up.
So this is how it started,right?
(07:38):
One of the thing I noticed whenI was in back office is that
people see each of your dressand want to analyze whether to
buy it, whether to replenish itin a different.
If it's a t-shirt, it's justsize and color, so you want to
see which color is moving fast,which size I need to order more,
which size I need to order morefor me, or so that makes sense,
whereas when you come to femaleathletic wear, then it becomes
(08:01):
more complicated.
You have cup size, you havetorso length, you haveor and the
regular sizes.
So now you are looking evenmore matrices to make your plan
to do analysis around what isselling fast, what is not
selling fast for a beginningseason, what I have to do.
So that's why the attributioncame into play.
That was one of the biggestreason people are moving In
(08:23):
traditional ERPs, even in likesee, if you want these custom
attribution, you need to pay fora software engineer, you need
to customize it, and when Ilooked at it, for me just felt
like an extension to MongoDB.
In typical database you think ofprimitives as numbers and your
alphanumeric parameters, butwhen I speak more to these
(08:45):
vertical, specific industriescolor, for example it's like a
primitive data type to them,barcodes, which you see at the
top it's like a primitive datatypes.
So what I ended up doing islike extending MongoDB's
functionalities to include theseprimitive types which each of
the brands, and from that webuilt a schema-less ERP.
That means we don't chargepeople for customization.
(09:07):
You come to us and say, hey,this is exactly how I look at my
business, these are theproperties I want to be tracking
for my clothes and we set themup.
That was compelling enough forthem.
They were willing to take therisk to do the data issues.
That's pretty much the reasonwhy people started and kind of
like it's important that whenyou are you, when you are in
(09:29):
your business, you get to makedecisions that makes your
business better and not changeyourself for the software that's
available, software that'scheaper.
That's first one.
Second one we did.
I started it as fully apparelspecific and even if you are not
in Shopify, we were building it.
But the more I started buildingit, the more I noticed is that,
(09:51):
unlike traditional ERP, wecould tightly couple with
Shopify.
Shopify has done a tremendousjob of exposing APIs and
whatever Shopify does well, Idon't want to repeat.
I just want to pass on thatload to Shopify.
What this meant at the end ofthe day for my customers is that
they end up being somethinglike a Shopify app.
Once you migrate to us foranother migration or if you want
(10:15):
to use another new software,you can use any of the software
in Shopify app ecosystem.
You don't want to.
Erp never becomes a bottleneckfor you to try different apps
marketing apps, offer apps,bundles, apps whereas with
traditional erp.
When you want a new app,because erp is not tightly
coupled with shopify, then yourerp needs to support those
(10:37):
integrations.
So that means you are notnimble, you are not fast and,
let's be honest, not every erpcan build an ecosystem that shop
has well, the amount of adsthat shoppers and that's the
other part of how I see that wecould build something that's
different while giving users thepower and bigger
functionalities.
John Kundtz (10:57):
So, in a nutshell, I think what you're trying to
say is you're sort of bringingtogether the I call it the back
office and the front officeright.
So the back office is all theERP, traditional inventory
management and orders and allthat stuff that goes in the back
end.
The buyer right never reallysees that.
They see the front end, whichis the Shopify, and so you're
(11:17):
bringing both of those together.
I think and that's probably themistake I'm hearing that maybe
organizations might make.
Harish Chandramowli (11:24):
Absolutely right.
Traditional ERPs used toreplicate what our data is in
shop.
If I ask them in the backoffice, that is disconnects and
everything we try to be like.
Let's bring things togetherbecause it's a single stack.
Why are you thinking as eachand everything is very different
?
John Kundtz (11:38):
Which sort of leads us into the next question we
were talking about in the prepmeeting was this whole idea of
workflow optimization a betterpath forward?
So what does that really meanin practice for, let's say, a
small or mid-sized organizationthat you might be working with?
Harish Chandramowli (11:55):
Yes.
So in terms of workflowoptimizations, a lot of people
have very custom workflows.
Some of them print labels Forsome of them, baha House prints
labels and each and every onelook, even the headcounts they
have.
The type of people they have asbusiness evolves is very
different, especially the biggerthe business, once you cross
that 10, 15 million mark.
(12:16):
Now you have an operationperson and you have a production
person who is different from anoperation person.
Now you have a customer serviceperson who is different from
these people.
How do you bring everyonetogether?
One of the things that a hasdone a really great job is being
able to build theseintegrations based on various
tooling.
You use various workflow.
You have the customizations atthe top, the last.
(12:39):
At the beginning, I usuallytell people to change their
workflow so that you adapt tothe software and be more
efficient.
But in the last six months toeight months especially when
people come and say, hey, thisis my workflow, I need this
customization I was able to goback and build it in two days,
just like the smaller changesand workflow optimizations, and
that's how I also see thingsevolving, moving forward and
(13:06):
being able to build these alittle more in a nimble, agile
and quicker way that you canactually go to people and say
this is my workflow, this is mybusiness context and engineers
come up with the software thatadopts to your workflow and your
business context and you don'thave to pay $100,000 to $100,000
for implementation.
John Kundtz (13:24):
That makes sense.
In the old days, you would tryto avoid customizations, correct
, because every time you did anupgrade, every time you made a
change, the custom integrationswere always what helped people
from moving from one version ofenterprise software to another.
It's not here that AI isallowing you to do that quickly.
Therefore, you can bring theboth worlds together.
(13:46):
You could quickly deploycustomizations that aren't going
to necessarily bite you in thebackside in the future when you
have to make upgrades or changes.
Is that sort of what you'resaying?
Yes, absolutely right.
So walk us through a real lifeexample of how that works.
Harish Chandramowli (14:04):
For me, the real life example is SKU number
generation.
Everyone generates SKU numberwith some kind of custom logic.
People were using spreadsheetto generate those SKU numbers
and I was able to just write aquick custom logic in the front
end that says hey, for thiscustomer, this is how they
create SKU numbers.
Like, if you have color red,that's number 48.
(14:25):
If you have that, if the cupsize is x, c, add c to n.
All those customizations now isso much easier that people don't
need to manually do it and youget your own specific things.
Number one.
Number two for example, if youwant people use project
management tools to createdesigns and everything, and the
moment everything gets approved,they go to shopify and then
(14:46):
they create.
The moment everything getsapproved, they go to Shopify and
then they create the product inShopify.
You don't need to do that AA.
You can build thatcustomization Once in my project
management tool.
I am approved directly, just goand create the product in
Shopify.
So those kind of biggercustomizations.
It used to take a week becauseyou need to know the APIs for
(15:07):
Shopify, you need to know theAPAs for these project
management tools.
Aa makes it easy.
Number two we always thinkabout product, as all the edge
cases, we need to fix somethings.
In fashion industry or in mostof the vertical specific
industry, people do the exactsame workflow.
You just need to make suretheir day-to-day workflow works
(15:29):
for these custom software andthat, so that means testing is
easier.
Users become your testers.
They're happy and you get allthese kinds of workflow
automations in a matter of days.
John Kundtz (15:41):
It leads me to the following question In your
experience, your clients or yourcustomers, do they understand
their workflow in the beginning,or is that part of the upfront
process I've been reading andlistening to and observing and
even experiencing is?
Ai is super powerful, but youhave to know what you're doing.
You have to really take thetime and understand what your
(16:04):
workflow is.
I mean, you can't automatesomething if you don't know what
it is.
Is that making sense?
Is that right?
Harish Chandramowli (16:11):
It's absolutely right.
It's one of the places that Iuse AI to meet myself more
better at sales.
So one of the things I do islike exercises I do in the first
call is ask them explain to mepurely in business terms, don't
explain to me how people useyour ERP.
I don't want to hear that frombusiness terms.
Explain me what people you haveand what is the functionality
(16:33):
for each one, and then I feed itto Android, which gives me a
really smart my JS workflowchart.
Then I sit with my customersand be like this is a very
visual workflow chart.
This is what I understood fromyou.
Can you validate?
This is how your businessoperates.
Then it becomes more easy tosay how my software fit into
their workflows.
John Kundtz (16:54):
So I love that for actually two reasons.
One is what you're describingto me is something I would call
you're co-creating with yourclient or prospect, so you're
trying to understand theirbusiness, probably better than
they may understand it I knowwhen I've done this.
A lot of times I've laid stuffout like that and the client
(17:15):
looks at me and goes, wow, Inever knew that before.
I don't even know that aboutourselves and I work here every
day and now it also allows youto follow up with them correct
and start to build thatrelationship.
One thing hasn't changed in thelast 40 years that I've been
selling is people still buy frompeople they trust, especially
(17:36):
if you're new, you're anentrepreneur, you've got a
startup, whatever, but itdoesn't matter.
You can be a large or smallcompany.
At the end of the day, they gotto look at you in the eye and
say, hey, Harish, I believe youcan help me, and you don't do
that just by sending them abunch of emails or just assuming
(17:57):
you know what they're doing.
I think that's great.
Let's circle back on the AIplay.
Ai is effectively being usedwhat you do without
over-promising what you can do.
I mean you described a littlebit about it in the pre-sales,
but over on the back end orthrough the course of the
buyer's journey and ultimately,the implementation of your
software buyer's journey and,ultimately, the implementation
(18:19):
of your software.
Harish Chandramowli (18:24):
So one of the things that was very
important, even as a CTO andfounder, is to keep an eye out
of which evolution of AI isproduction ready versus which is
not.
We hear a lot about agents, howagents can do everything for
you in production, but think ofsomeone getting an invoice,
keying the invoice in to makethe payment.
If an agent is doing it, if itmakes a mistake, then the
repercussions are big.
You are not paying the rightamount, your books doesn't tally
(18:46):
, then you need to go throughlike thousands of invoices to do
it.
But so agents is not there forthat specific use case, whereas
the use case where, example,some people in our industry miss
emails that says purchase orderis getting delayed, I could
easily parse every email thatcomes in.
When I feel something seems tobe off for their manufacturing
(19:09):
communication, I can flag thatemail at the top so that people
can go and verify it.
And a is that a can do thatreally well.
So understanding which part ishype and it and AI is that AI
can do that really well.
So understanding which part ishigh and which is production
ready is really important andwhich is very critical to users
is important.
The second part isrepeatability.
Right, lot of times when I talkabout invoice passing or
(19:30):
uploading your customer ordersto understand how they operate,
some of them needs to berepeatable and you need to be
sure that when you repeat it itwill give you the exact same
results.
And AI doesn't do that.
But what AI can do is if I useAI to build those customizations
as an app, if I do it to buildall the custom logics, then you
(19:51):
can test it, deploy those customlogics and productions.
You know that when you areusing the product it's not AIs
they'll go and give you theexact same results, whether it's
a SKU number generation, like Imentioned, or creating some
custom PDF to send to yourcustomers for invoicing All
those things.
Once you use AI to build itfaster, you can use it in
(20:12):
production Agents.
Yes, email passing agents exist.
If you come up with a invoiceor, like we call it thing, list
in our industry which comesmanufacturers and what item they
are producing and how much, itused to be hundreds of lines.
Someone manually used to enterthose key in those things in erp
, but now you can upload it.
(20:32):
We pre-pass but we don't reallyenter the value.
Rather we ask someone to verifyit.
Now your job of keying in forone hour changes to a job of
five minutes of verifyingwhether the past values are
right.
So that's how I see ai, andobviously a month from now, like
you mentioned, if you ask me,maybe agents is there, I would
say agent would do the wholething.
(20:53):
It's all the hype.
You can sell a dream, but youalso need to be understanding
which is production ready, whichcan be used in real life today,
and not just what is going tocome.
John Kundtz (21:03):
Yeah, I'm glad you hit on agents, because it drives
me crazy, because if you watchthe external marketing and
newscasts or whatever,everybody's talking about agents
and, as you mentioned, it'scoming, but it's not the panacea
, especially today.
So great advice.
So, speaking of advice, tell us, as a founder, if there are any
founders listening today, whatadvice would you give to them
(21:25):
about building the rightoperational stack, particularly
when it comes to technical?
We're all overwhelmed with techoptions these days.
Harish Chandramowli (21:33):
Keep it simple.
Pick things that you are reallycomfortable with.
I even while working atmongolian stuff, while I have
seen people how they usedatabases, my first thought is
that unless you are like thecoinbase and google of the world
, the reality is it doesn'tmatter, even if your query is 10
milliseconds slower, your userswon't do it.
So pick something that you arecomfortable.
(21:55):
I mean obviously.
Obviously there's other changesif you're building a product,
but when you are building theseinnovations in ERP or these kind
of like very real-worldproblems, trying out different
ways, trying out differentthings helps.
People needs a lot ofvibrations.
So pick something that's soeasy to you that you can trade
(22:16):
until you see a market fit.
Then, obviously, you can justgo back and upgrade everything.
John Kundtz (22:21):
That's great advice , I agree.
I see people, especially in thetechnology space, just trying
to create solutions purely onthe technology, and what I
really have picked up throughthe course of this conversation
is you're really focusing onwhich is what I advise all my
salespeople to do and all thetechnical people that I advise
(22:42):
the product managers of theworld and stuff like that is
really focus on the why.
Why is what you're buildingimportant to, ultimately, your
end user or your buyer or yourcustomer or your client?
And then build it quickly andsimply and then iterate over and
over and over again.
Harish Chandramowli (23:04):
Yes, One example I would usually give
people is like when I talk aboutemail passing, I can make sure
Google can tell me when a newemail arrives, or I can just
query every 10 minutes onwhether this person has new
emails.
The later one is quick to build, not the most efficient way,
but until you prove out thatit's useful, just in the
quickest way.
John Kundtz (23:24):
That's great.
Yeah, it's a classic agilemethodology or mantras I like to
say fail fast and cheaply.
Right.
So build something.
Test it out.
If it doesn't work, throw itout.
Pivot If it works.
Great.
If it needs to be better, testit out.
If it doesn't work, throw itout.
Pivot If it works.
Great.
If it needs to be better, makeit better.
All right, before we wrap up,switch gear.
What's something you've learnedthe hard way in your experience
(23:45):
that has shaped you to run yourbusiness today?
Harish Chandramowli (23:49):
Personally I feel when I joined Hopkins I
had to take two jobs and likethe five grad courses that means
I kind of got so used toworking long hours and working
long hours is very, veryimportant as a founder at the
beginning stages because youneed to find the product market
(24:12):
fit quickly.
So that experience definitelyshaped that.
And over the last two yearsthings I learned as founder is
not saying yes to all thecustomers and being very
pinpointed in the type ofcustomers you want to onboard at
the beginning.
It's so exciting to get lost.
I got lost where some peopleknow who are not in shop.
(24:35):
If I have my, I love thesolution I want to onboard and
end up building a lot for them.
But looking back, no, I don'thave the resources.
I don't have the money beingleased and focused on a small
group, even though you can havea bigger vision, so important I
think that's great advice.
John Kundtz (24:50):
I've seen that mistake that you said you
avoided over and over again.
Where, especially as a newcompany or a founder is every
prospect is a good deal, ittakes a maturity level to say no
, we're not a good fit right.
And understand that you couldget just sucked down a rat hole
(25:11):
and blow all your cash on oneclient and therefore have so
many small early stage companies.
It's not that they don't havegood ideas.
It doesn't mean they have badmanagement.
It just a lot of times theyjust run out of cash and if you
don't have cash, unfortunatelyall the good things you do can't
happen.
All right, we'll wrap it up.
(25:33):
Want to thank you for sharingall these insights and
experiences.
How can people learn more aboutyou and your company and your
services?
Where do they go?
What are your socials?
What do you want to share withthem?
Harish Chandramowli (25:47):
I always tell people they can follow me
on LinkedIn.
I usually update, keep myLinkedIn updated.
You will see what we are doingin the company in my LinkedIn.
John Kundtz (25:55):
any thoughts I have on LinkedIn and if you have any
questions, you will see what weare doing in the company in my
LinkedIn, any thoughts I have onLinkedIn and if you have any
questions, you can obviously DMme and always happy to reply and
of course we will include thoselinks in the show notes so you
can check it out.
You want to learn more orconnect?
And, harish, I'll give you thelast word before we wrap up the
show.
Harish Chandramowli (26:15):
I'll give you the last word before we wrap
up the show.
One of the things I would sayis just have fun.
If you're picking a software,any job, whatever it is, just
say pretty much B-Bull, that'sawesome.
John Kundtz (26:26):
All right, bud.
Hey, I'm John Kuntz.
Thanks for joining us on thisedition of the Disruptor Podcast
.
Have a great day.
Thanks a lot.
Thank you.
Popular Podcasts
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.
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
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.