#544 Reinventing Retail OS: Harish Chandramowli on AI, Workflows, and the Future of Fashion Tech

[00:00:00] 

Mehmet: Hello and welcome back to a opposite of the CT O Show with Mead today. I'm very pleased joining me Harish. He is the head of AI at Good Day Software. Uh, usually Harish the way I like to do it, I don't like to steal from my guest [00:01:00] spotlight, so I keep it to them to use themselves. So tell us a more about you, your background, your journey, and what you're currently up to.

But just as a hint to my audience, we're gonna talk, you know, about something. I didn't cover it much, honestly. Um. It's about, you know, the power of AI and technology in, in fashion, in retail. Right. And you know how this can maybe open a discussion with Harish today regarding that aspect. Of course, we're gonna talk about AI and how Harish he, he looks, you know, to, to the, to the industry that are further ado.

Uh, Harish, I will keep the floor to you. So let us about yourself. 

Harish: Hey, it's a pleasure to be here. I'm Harish. I have always been in technology sector working on a bunch of data related products, just in terms of my growth or failure. It started at Johns Hopkins. I went to Johns Hopkins for cybersecurity.

And when I wanted to look for a job, I realized I always liked building products. [00:02:00] So I ended up not going into this consultancy part of the security rather end up joining Bloomberg as a in the security engineering team that led me to join MongoDB as one of the early cloud security hire. Mm-hmm. And it has been a great drive.

That company expanded very quickly. We got a lot of users. That means you get to learn how to build a product that's very scalable. And I think around like four or five years Mark, I felt like MongoDB has also grown and I love working in much smaller companies. Mm-hmm. And I also felt like I've seen like retail being one of the biggest consumer of MongoDB while I was on call and everything that let me to be just curious around what that is.

And I kind of like went to a store called O in New York. Sat in their back office, literally just like saw how they operate, sat in their meetings. What stood out to me is that whether it's retail or any e-comm or pretty much [00:03:00] any business, your back office operation. Mm-hmm. Still the data and workflow problem, those were the major two challenges for any business to run effectively, especially e-comm businesses.

That led me to start FLA as a city board. Two years we are building FLA and at some point, yeah. Most startups leave at this inflection point where we need to decide whether to keep going with it, do a more fundraise, or it's kind of like we are not where we need to be. And that kind of led me to join Good Day Software as the AI head of ai, purely because now I get to concentrate more on the fancy new AI part of like how things works, how we can help customers in retail industry while not having to worry about the business aspect of it.

Great in the sense like my CEO and CPO are like, all those people are really good. They will run with it. 

Mehmet: Yeah. Yeah, that's, that's a fascinating, uh, journey. Harish really, you know, [00:04:00] um, you know, and I'm sure you gained a lot of insights. Now you kind of mentioned it like, but very quickly, but I'm curious, you know.

Uh, and I know like, uh, uh, when I was researching it, probably I saw it on, on your LinkedIn or, or, or somewhere else. You mentioned like you weren't a, a fashion person or retail business, uh, per person by training. Uh, yet you identified these, you know, pains that you just mentioned. So tell me how your previous experience in data security engineering help you to spot this opportunity that you can come and fix.

You know, this, this domain. 

Harish: Yeah, so I mean in none of, in all those podcasts I didn't have spoke much about the technical aspect. So I'm kind of like having the technical aspect of where it came from. So I was in this store, right, and I was operating and I was also speaking to the bunch of like fashion art brand who are like female fashion art brands because obviously there are more female fashion [00:05:00] brands than men men's out there.

There are few things that stood out to me at that point. One you like. T-shirts are colorant size. If you are seeing brawler, it would end up being like a cup size to length, color, and regular sizes for different parameters. And when people are in the back office and they're talking about data and how they are like seeing sales, how they're seeing their purchase orders, it end up being like, I'm seeing color and barcodes.

And how the existing software works, especially in the SMD market, is that you have these ERP, which are very rigid when you talk about should they say option one, red. Option two is size. And then if you're talking about some other properties called option 3, 4 5, and then when you want to run these reports, people take the data out of the reality and they were trying to stop in extra properties like materials or locations or things.

Run an [00:06:00] ETL pipeline, feed it to some kind of like a data lake, and then they were running reports coming from MongoDB background. For me, it's like why have these multi-step process exist? People who create the products know the extra property they need it. So it should be a very unstructured ways where when you create the product, you can say, this is a T-shirt, this is a cotton T-shirt, or whatever you want, and then run the report directly.

I saw this as an extension of MongoDB that. Text and numbers were the primitives when we were building the serverless architecture, whereas in fashion or e-comm, the primitives they use is called color, which is a combination of Pantone and bunch of different properties. So how can I extend the basic MongoDB functionality with all these custom business data types, and then let users customize their ERP from the get go.

Rather than having to have like engineer customize it rather than having to have a separate software engineer running an ETL pipeline to be able to [00:07:00] generate a report where they can see these extra properties combined with your sales order. 

Mehmet: That's, you know, I, I like this, you know, it's exactly. You know what, how I love usually people who works in technology to, to, to be thinking it's about, you know, bridging, uh, the business strategy, uh, and domain with how we can, um, you know, get that, uh, from technology perspective.

Right. Um, I know one of the things that, you know. If you want a little bit, dig into the technology itself. So one of the themes, uh, I know you've spoken about is, uh, data integrity and messy workflows. Uh, can you little bit, you know, tell us about this. Like, I know for example, you said spreadsheets, secretly killing fashion brands, profitability, like unpack, like how big is the problem and why do you think some brands they underestimate that?[00:08:00] 

So at the beginning, 

Harish: people. Don't have money. E-comm business itself is really brutal. The more you learn about the business, the margins are low, so every expense matters. So people don't want to put money investing in like a bigger software and to be fat to them next week of the world cost you a hundred, $200,000 a year to just implement.

So it is pretty expensive. So people fall back to spreadsheet. Now the challenge becomes how to coordinate between different teams, how to keep different spreadsheet updated, how to know what is happening even between your business. Right. An example I can take is a purchase order. Mm-hmm. Manufacturing typically takes around six months to place an order for it to get shaped for it to like come to your warehouse and people book models for photo shoots so that when the t-shirts or anything are right, you can take the photo shoot, put it in the website, start marketing.

But there are also delays that happens whether it's a tariff, whether it's a vault, [00:09:00] it doesn't matter. And if you are purchase order manager who is solely responsible for procurement, do not communicate to your man marketing team that there is a delay. They would've booked the models. We are paying for the models, but the item hasn't right in that it's just a small one.

Use case of example. Now, if you think about like various departments that happens behind the scenes. To run a company and they don't interact with each other. They have spreadsheets with same data replicated in 10 different places that's not getting updated. Now you are losing money. Mm-hmm. So it's a matter of lost revenue than just like, oh, I forgot to tell someone.

Right. And especially, like I said, e-comm industries, brutal cash cashflow, brutal industry. And you overpay for things. You are out of business in three months. 

Mehmet: Right, right. I 

Harish: can't solely say people don't think about it, right? The challenge here is the fact that softwares don't make it [00:10:00] easy for them to enter those data.

So existing softwares don't make it easy for them to feel like, feel, they don't feel empowered with existing software, that they don't, they feel that it's a pain to go and enter in data. Example for me is that packing is, which comes from the factory, has hundred rows and no. Very ERPs will make it easy to enter those hundred rows.

You literally have to key in and it's a boring book, so they want to wait until factory confirms. This is the data that's coming in before they enter the data into an ERP. At the time you are losing the change set. You are losing the changes that comes out. Evolution of the ERP, like negotiating cost, what, how we negotiated the lifecycle of the purchase order.

Everything is lost. That's clearly because EIP was not making it easy for them to enter the data, and that's where I feel like making the workflow automation so easy would help people be in the software, [00:11:00] happy to use the software and get most out of the software. 

Mehmet: Right. Uh, I love this in depth understanding of the problem Harish, and, and, and how you're tackling it.

And, you know, of course, like understanding the pain and, you know, like I tell people usually in any, any domain, like if we understand. The, the, I would say the negative outcomes of doing nothing or like the cost of doing nothing like this is when really our mindset will start to flip and we start, oh, like, yeah, this, this makes, uh, a lot of, uh, a lot of sense.

Now, you know, when, when I go to Good Day software website, I see, you know, the first thing that. I'm, I'm, you know, seeing on my screen, seeing the first AI powered and then ERP is like striped out retail os right? Built for, for brands. Now my curiosity as, as a a, a technologist back in the [00:12:00] today is how you define the role of AI in this context, right?

And why, why the ERP is striped out. 

Harish: The one, what is ERP? Just to give a. Highlight of people who do not know is that it lives in the heart of the business. Mm-hmm. So most business users like 30, 40 plus softwares sometimes just to run it. Example would be like they use something called Shopify for your websites for product and selling That is something called.

PLM, which is like production lifecycle management, where designers can do different things. You can go back and forth about it and people use a different software called New order or different things for wholesale. So what ERP does is bring all these platform together, and the reason why I think of it as a voice is because at the end of the day, how you bring these things together, changes with ai.

[00:13:00] Imagine what I was talking about. It takes a hundred thousand, $200,000 to implement ERP. Why it takes that much money? Because previously it used to be rest a PA. You need to hire engineers who specifically implement integrations with ERP and every software you use. But the world has changed. CPS and all those things make it so easy to bring together these software.

That's why it kind of like comes to the type of retail OAS versus an ERP. Then after return, doesn't matter. What matters is that one you communicate with your factory through email and everything happens in email, and an ERP can just grab the details from email and put it in the system. You can use a hundred of of software.

We don't care. We will make sure integration is so seamless that everything goes through ERP. Your core data lives in ERP. You can get your snapshot of your business when it's new items [00:14:00] coming in, how much I am selling, how to plan the new season or new year, what to purchase good day as ERP would act as a source of truth.

Mehmet: Right now, uh, going into the, the, the AI part also, uh, Harish with you? Um, like what kind of, you know, so for me, when, when, when someone tells me I have something AI in the solution, like, is it like the, the one that, uh, you know, it's just like the chat. Based one, the conservative one, generative ai or like agent based, like, you know, I'm, I'm just trying to understand the context of, of the AI here.

Harish: Oh, that's a perfect question. Some of them I'm going to chart in the product. Some of them is, uh, vision. But you are pretty much spot on, right? Mm. One of the things that [00:15:00] stood out to me, especially like with Goodday when I was looking at it as well, is like, AI is not just it. Chat bot chat and for a long time people always had AI as a chat bot, which is a easier thing to build.

I a database, go and grab and answer the question for my customers, which has always been like a traditional one. But when it comes to an operating system or these back office operation agents play a huge role. So this is a combination of both agent teeth and kind of like, some of them are just enablement of small things.

Let me give an example of both the worlds right one. I talked about packing list, which has a hundred lines of data, comes from a factory and it has like so many different skews, different insights that people that keying into the ERP. Now with ai, upload the pdf F that you get from factories. We will pass it.

We will prefill the values inside your work. We call it packing list as basically a CSV that [00:16:00] gets converted to a table. Mm-hmm. And a can do it rather than like you keying in like thousand lines or a hundred lines, you just verify it. And that's not an agent. Agent as in like it's not working on its own, but it's also not a chat bot.

It just makes your life easy. Imagine previously it took each one two hours to key in this data. Now it's a matter of 10 minutes. You upload, everything gets prefilled. You, you, you have 10 minutes to just be like, is my invoice values match as the one that AA passed, if not directed, undo. So those type of things is to help user be more efficient.

Mm-hmm. And then the other part is agent tick flow itself. If you get an email, all these POS comes in an attachment. Right. So the a previously, you look at the email, you need to pass, download the attachment, and then you go to your ERP and do stuff. Now with a doing those integrations, workflow automations becomes so easy.

So you can set up saying that, Hey, if I receive a email [00:17:00] from my vendor, can AI look and flag if it's a purchase order, if it's a sales order, and if it's a purchase or sales order, can a automatically put that into an ERP so that when I log into the ERP, I just verify. Oh, I got an email. It looks like a purchase order.

This is the data that needs to get into my system. I'm done. And those are agent flows. Mm-hmm. Other agent flows includes simple things like, I mean, it's not simple. People spend hours every day that, Hey, my vow says I have 30 hand handbags, but my e-comm site has only 20 handbags. That is the discrepancies.

Where am I losing that? Count. Sometimes it would be like, oh, you got 10 new handbags from your factory that the virus hasn't encountered, or it hasn't, but people spend so much time on it now. Agent can do it. Agent can grab your inventory data from different places, compare it and flag it so that you don't spend hours trying to reconcile what is happening.

So those ends up being an agent flow. 

Mehmet: Got you, got you. No, this [00:18:00] is very, very informative, I would say. And you know exactly what, what I was, uh, thinking about Harish. Now I might, you know, I'm trying to, to, to position this question and feel free to, to ask me if it's not clear. Yeah. Uh, it's kinda not directly related to the tech itself.

Right. But it's related to what you're doing currently. So. And of course, given your engineering background, you work with MongoDB also, like on, on a, you know, which is a database, uh, company. So, uh, now, when, when, when these brands come, right? Um, so. Do you think the data and you know, the workflow would be able also to, so, so let me, let me ask the question in a different way.

So I think, you know, 'cause this has just popped in my head. I'm trying to get my thoughts together now. Uh, there are different markets. So there's the US [00:19:00] market, uh, and North America market like Europe, Asia and so on. So let's say. I'm now here in Dubai, right? And I want to leverage, you know, known data, which is, might be like available data from like maybe statistics perspective, what workflow work and so on to develop a new brand in, let's say in the emerging markets, right?

Like do you think. The data that usually maybe a model would have, or maybe like some, uh, workflow maturity that would have, can be localized for me. You know, when, when I'm trying to, to find, you know, something that fits me. So if I bring, for example, a, a, a data engineer as, as. New brand, uh, entrepreneur. So how, what, I'm trying to understand how much, you know, I would be able to leverage, you know, some, some of the models probably in that domain for me to start if, if I, [00:20:00] the, I hope, I hope, trust.

Yeah. Yes. 

Harish: So most of the time I can talk from e-comm industry perspective, it's all the same, but there are nuanced differences, which is pretty much like dollar versus a RU piece versus a. Right? 

Mehmet: Mm-hmm. 

Harish: So that needs to get normalized. There are a few changes that would come in because, so that like as you expand your business, your profit margins are calculated correctly, repeat conversions or dollar conversions.

Pound conversions are somewhat like configured and set, set it in your way so that you get a clear view on how much is your profit. Mm-hmm. So there are obviously nuances in that, but overall operation wise, it's pretty much the same. So the software tend be used across the countries, but you tend to, you tend to start with one market, not because ERP can't work across different places or the data modeling and the AA training won't work across the places, but the software [00:21:00] people use very, and example is like European e-comm that are like BigCommerce and other stuff who are bigger and you want ERP to integrate with it and grab data.

Whereas in us, Shopify is the biggest player when it comes to e-comm front end. So ERPs often have to integrate with that. So when you think about expanding market is there are a lot of like pieces around the core model, core, a training that changes based on the country you are going to be in. But the core model, core program that you create is going to be the same.

Another example for me is accounting, right? Like the accounting piece, which isn't for us. Customers that you are building might not exactly apply to other countries because it becomes country specific operations. But business itself, we don't think differently, right? I need to know the operation operating.

It's funny that one of the terms people use here is like [00:22:00] operational accounting versus business like country's accounting, operational accounting. It's like something you want to be more close to real. You don't care about the one sentence, 2 cents that's missing 'cause you just want the bigger one.

Whereas real one has like more nuances that's associated with it. Real one makes country specific thing. Operational, accounting, operational, they of looking at your business training is common. 

Mehmet: Got it. Yeah, no, exactly. You know, and, and thank you for, you know, being able to understand, because I'm trying to explain it also from, for, for like in layman terms, so maybe people who aren't familiar.

So, but, but you know, you helped me and thank you Harish for this. Now if, if I want to look at, you know, again, talking from the AI perspective, so we are seeing more and more, and you know, this is one of the thing that. Uh, using, you know, AI myself, but they signal that I, I, I spotted and I expected it, uh, is like the [00:23:00] rise of this class of technology, which is like AI, verti, uh, you know, verticalized ai.

So an ai which is like. Exactly for a specific vertical moving forward ish. From your experience, do you think like we're gonna see more and more now? For example, today we are, we are discussing, uh, um, you know, this in, in the, uh, fashion retail, uh, perspective. Are we expecting to see like more rise of the specialized AI in specific vertical?

And in this case, what would be the mo in your experience, like how we can, uh, as founders and, and you know, entrepreneurs claim like, hey, like we, we, we are better than others, so, so what's your take on this?

Harish: Candidly, I feel no one knows what, where all this is going to go. Right, because the answer I would've given a year back is so different from the answer I am going to give right now. [00:24:00] Uh, things change very quickly. So I'm going to come with an saying that, hey, if you're listening just six months from now, I even, I would be giving a different answer.

Uh, the day I think about it is one, the horizontal aspect. I feel like more and more open AI and like ropy of the world are trying to take over. Example, when I talk about workflow automation six months back, a year back, the way I thought about it is that I need to have a builder. Salesforce has a builder for CRM.

Everyone has it. So I need to come up with my own builder for an ERP value. Drag and drop. Build your workflows. And some of them are automated. Some of them might be a small piece of code that runs, and that's how AI needs to bring workflow together. But if you ask me now, agent Kate and agents of the world has done so much automations.

I only care about doing the vertical aspect, which you are talking about, that I [00:25:00] know the data I have, I know the data I have collected from my customers. How can it make it more useful to people? Example. Example is inventory movement tracking. When an inventory comes to your battles, when it goes to a customer, and if someone does returns when it comes back, what value is coming back?

All those are really hard data to collect and charge. P and pics of the world don't have those data 

Mehmet: right in 

Harish: the exact way that's needed. And we collect those data and we can make sure we can reuse the models. Publicly from these like big lms, and then combine that with the data, train that with the data to help people understand or make sense out of the data, right?

For example, if you are hiring a new inventory manager previously, you need to train them where to go and look for data. What data to look for. How to make sense of your business. Now, if you're hiring a new [00:26:00] inventory manager, you can just be like, Hey, go and ask this chat bot about our previous decisions.

What has been done, where inventory is and chat bot would give you the answers, 

Mehmet: right? That 

Harish: gives the AA more of like a, and know this cannot be taken over by the trillion dollar companies out there because it's very unique in my industry. I know how to train this. And I also own the data, not just the training model of it.

Mehmet: This is fantastic. Uh, and by the way, just, you know, Harry, I, I'm not sure if you have seen like how I'm asking you. I didn't tell you, for example, tell me in two years, five years, you know, since, uh, you know, I'm not sure if this is, uh, a luck or not, but, uh, I started the podcast about the same time when Chad g PT came out to the world.

Like, it was like only two months maybe. And, you know. I spotted like also something that, okay, things are gonna change very fast, so I don't have to ask this question. I try to see like maybe directions, maybe I, I would say like, Hey, what, what people will [00:27:00] hype about. But to your point about the data, and I'm asking you again because of your background now, one of the biggest, uh.

Uh, topics, you know, that, that people talk when, when we discuss AI is about training data. Now, I think, and I believe, correct me if I'm wrong, you know, in your domain is not hard. Like you, like there is no scarcity of data always to come in to train the model, right? Because, you know, in some other domains, what I've heard from people, like, yeah, like they're gonna run out of the data and what they have to do is just to go and find like a.

You know, synthesized data, which can be used actually to train the model. Uh, like can you like, just clarify this? 

Harish: Uh, yes. In this case, data are real. It's not synthesized. You don't have to extrapolate things. I know if you're in like a biotech or something off and it's like maybe we don't have lot of data, we can use a to extrapolate and then like run some models.

That's not the [00:28:00] case. These are real data that's coming from a real world physical operations. 

Mehmet: And you know, we will not get out of data any soon because people are always, you know, doing transactions. Right? Yes. 

Harish: I think we won't get out of data very soon. What I am noticing with e-comm industry, which I'm curious to see how things always, where the data comes from, right On, how it is being used.

For example, you would put ads in Google search so that you can drive business to your e-com. Now Shopify has enabled you purchasing in charge GPT. So the behavior has changed where you might ask charge g, PT like, I am looking for X, Y, Z, I want to order an order within charge GPT. 

Mehmet: Mm-hmm. So 

Harish: now where the data is being used, how the data is being used, changes drastically.

So now you are not looking at making a good ad. Now we are thinking through like when people ask questions, how do I enrich the data so that like AI can [00:29:00] surface my product? 

Mehmet: Right. Yeah. 

Harish: So those, those things are going to change though. Data itself is going to be there, but how it is being consumed and what data is going to be consumed is going to evolve.

Mehmet: Absolutely. Yeah. Yeah. There is no, there is no doubt about it. And, and spot on. Now, Harish, you know, one of the things, um, I, I'm curious about, um. Especially, you know, you, you've had this long experience and I'm sure like maybe some of, of the audience would be thinking like, um, like Harish is so, you know, uh, you know, this is what we call, you have a great business acumen.

Uh, usually business people, sorry, technical people. They, they, they kind of, especially when they start, you know, their own companies or for sometimes they co-found a company with someone, they have this issue, uh, that they are new to business. So as someone who, you know, [00:30:00] transitioned, you know, to, to, to have this full vision of, of the both world, like.

What mindset? I would, I would say like help you to, to also have this shift for you. And also like what kind of maybe resources or like what, what kind of, of, uh, of things they have to go and learn so they will be able to be comfortable to talk business language that could that be sales, marketing, GTM and so on.

Uh, 

Harish: curiosity. 

Mehmet: Mm-hmm. 

Harish: I guess that's the main thing. And the context is like, I was sad because at Mongo I was, no, I had no, I was not in like a very customer facing role, right? Mm-hmm. Obviously, I am involved when things go down. If top one person's database goes down, I'm involved. Yes. I'm very technical, like I have patents.

We do research, we like MongoDB is a technical product, 

Mehmet: right? 

Harish: And moving to fashion, which is. [00:31:00] Exactly the other end, but also even Bloomberg, right? I was in security, so my customers are software developers within Bloomberg, right? So it's kind of like a very different world. Initially I would say it was a little harder, but Charge JP has made it so easy.

So my customers directly interact with me and. The way I do it is like sometimes I just put like, Hey, this is the reason why something works. Something doesn't work into charge JPD. I'm like, I'm a software engineer. This is a technical aspect of it. Can you kind of like change it up so that I can communicate this information to a totally, totally non-technical fashion creative personality and chat will check out some answers.

I use that as a baseline on how to communicate to people. 

Mehmet: Yeah, 

Harish: and it helped a lot when it comes to sales, marketing and stuff. Someone who doesn't know anything about it, who is not very comfortable sitting in a sales call tragically helped me a lot. 

Mehmet: Cool. [00:32:00] 

Harish: I believe other aspects, 

Mehmet: yeah, please. 

Harish: Yeah, please.

Other aspect of learning the business itself. I also a pretty comfortable person at MongoDB, right. The company grew, I grew with the company. It's all like. Nice. What made me leave a stable job for a startup and founder is again, the C aspect of it. I was in these like fashion shows. I spoke with fashion people and when they talked about the challenges they have, it just thought that like it's a data problem.

I love how to make use of data. It's something I feel so passionate about. And I want to do it. That's it. There is no, you know, there is no right or wrong answer. Just like I felt passionate and knowing your customers is the important thing, being that sometimes literally just counting things in their back office, being in the meetings with them, in the back office, just observing every piece of their operation helps you learn.[00:33:00] 

Mehmet: Great. Yeah. So, so I like this. Like there is no, uh, right or wrong, right? So, um. Uh, this is, this is by itself, it's a mindset Harish, like also to go and explore and, and try new things. Curiosity, which you, you mentioned also as well. Um, absolutely. Now I gotta ask, it's not an odd question, but, you know, uh, part of, uh.

You know, of the audience that comes and listen to the, to the podcast or like sometimes ecosystem builders and sometimes they are investors themselves. And you know, this, uh, the thing we discussed, uh, outside of the podcast, like, uh, you know, I, people sometime ask me, um, there's a lot of, you know, and we said like, AI is, is mainstream now, right?

So we, we could not. Skip it. You know, we cannot, we cannot, uh, don't talk about it. So, uh, but I want you to, to, for, with all the experience that you have. So imagine someone brings you, you know, tells you a pitch or maybe, [00:34:00] you know, like they, they describe what they are building. Uh, so what would be a red flag for you?

Like, like should something like, uh, no, there is something not, not fit here. What that would be. 

Harish: So when I was switching, one of the feedback I get from investors is, I'm too practical. For me. Red reflect reflects me different for different people, right? When someone says, AI is magical, this is going to take over everything.

I'm going to replace entire company with ai with this product. You know, oh, that's not happening, right? No. So. B from an engineer. When I hear the pitch, I am more curious around when you pitch, how are you handling things that go wrong? 

Mehmet: Mm-hmm. An 

Harish: example I would take is that I could make an AA that is going to look at your invoice, automate your payment to your vendor, and not [00:35:00] have humans involved and say, AA makes a mistake and pays the wrong amount.

How are you going to flag it and how do you even reverse the transaction? So right. Have we thought through that to apply a near business, to make people efficient and where not to in these cases, it's as simple as like, Hey, I will give, rather than users entering the data, I would stage all the data and once they approve, I will make the payment and this is going to make someone efficient.

It's a good pitch if someone says, behavior seems you don't need an account, and the payment will automatically happen. Like. No, at least not enforceable future because, uh, error handling what happens when things go wrong is scary. 

Mehmet: Yes. That's 

Harish: how I look at things. I'm like, are you thinking about real world implications when things don't work?

Mehmet: Uh,

you know, this is exactly the way I also tell people like. [00:36:00] We, we, we, we need to, and you mentioned, by the way, Harish yourself at the beginning. Um, maybe not directly, but in an indirect way. So I need to be able to describe the, the, the, the, the customer journey rather than, you know, just focusing. On, on, you know, and you mentioned about the pain.

The pain is big. Like, for example, they're losing money. They, they're not aware of this. Right. Uh, so you need to focus on rather than, as you said, like saying, someone tells me, yeah, yeah. Like, I'm, I'm, I'm replacing thousand people with one agent. And by the way, maybe, maybe, you know, sometimes, uh, uh, in some domains.

Uh, forget about investors from, from, you know, go to market perspective. If you go and say this, you are gonna create enemies on the spot, so, right, because you're, you're literally telling someone sitting there, Hey, I'm here, I'm here bringing something to, to, to, to replace [00:37:00] you. And yeah, there will be no need.

So you need to tailor the message in a way. Um. Yeah. It depends on the domain, but, but exactly what you said. Yes. Right. Uh, Harish as we are almost, you know, uh, coming to an end. Um, like final thoughts you want to share with us and, uh, where people can get in touch with you. 

Harish: LinkedIn, I am always available in LinkedIn, and my LinkedIn ID would be Santa Clara Harish se Harish.

Uh, file message. Just be curious. I think we are at the moment, especially as someone from a technical background, that everyone is in level playing field. No one knows what, how aga will be applied in the world. There is no secret to it. There is no one who is smarter than the other person. We are all in the playing field, which is so interesting.

Time to explore new things, come up with new [00:38:00] ideas, experiment different stuff. There's going to be thousands of pet dot coms. There was also one or two amazon.com out there from.com, and that's going to be the same when AA is, when a new technology comes in, we are all in level playing field, so just let's just have fun.

Mehmet: Yeah. Yeah. Um, although like I, I hope, I hope Harish, like people learn from, from the mistakes of the.com bubble, like this question, uh, keeps, you know, as you said, you know, I tell people I don't have a crystal ball, but I, I say. Humanity history shows us that we as humans, we, uh, we learn from previous mistakes.

Yeah, we keep doing mistakes, which is not wrong. Don't get me wrong. We gonna, we're gonna always make mistakes, but I mean, we'll not do the same mistakes, maybe probably new mistakes. So, uh, I'm, I'm, I'm bullish myself on, on ai. You know, what you're doing is fast and fantastic at, [00:39:00] at, uh, good Day software. So, uh, because it's also empowering.

Uh, you know, brands and sometimes these brands maybe are created by entrepreneurs, right? So, uh, so, so this would help people to succeed at anything that help people to succeed. We get excited about it. Now, regarding the links you mentioned ish, all they will be available in the show notes. So if people are listening to us, uh, they will, they will be able to find, of course, you'll link it in profile also as well.

Uh, and again, I, I want to thank you a lot. I know how busy it can be. Uh, so you took the time to speak to me today. I really appreciate that. Uh, and this is how I end my episode. This is for the audience. Uh, if you just discovered our podcast by luck, thank you for passing by. I hope you enjoyed it. If you did, so, give me a favor, subscribe and share it with your friends and colleagues.

And if you are one of the people who keeps again and again, thank you for the support. Thank you. Thank you very much for the laity. Thank you for pushing us up to. The top, top 200, uh, charts in the Apple Podcast, uh, platform across [00:40:00] multiple countries. So I really appreciate that and I'm really, really, uh, you know, feeling humbled and, and and grateful for my audience.

Thank you very much, and as I say, always stay tuned. We are gonna have a new episode very soon. Thank you. Bye-bye.