Podcast Intro 6: you
raspberrymusic tech logo reveal short hi tech intro 414307: you
Robert Curran: Hello and welcome to Talking Transformation, the Appledore Research Podcast. My name is Robert Curran, consulting analyst with Appledor. Today we're going inside one of the most sensitive yet critical programs that any telco can make, transforming its core commerce systems. So joining me today to unpack how operators are navigating this challenge. I have John Abraham, principal analyst, Appledore Research.
John Abraham: Hey Robert.
Robert Curran: Hitesh Kumar, SVP for consumer technology and digital innovation at Indosat Ooredoo Hutchison.
Hitesh Kumar: Hey.
Robert Curran: And Saurabh Sharma, I C T program Director for IOH for Ericsson.
Saurabh Sharma: Hello Robert.
Robert Curran: Welcome, welcome everybody. John, You recently published some extensive research on telco commerce and B2C systems. What's driving this sudden executive urgency?
John Abraham: what we have seen is especially over the past six, seven years, and particularly post-COVID, there's a lot of focus on transformation of the customer interaction model, especially via digital channels, and to the extent that this has become a boardroom priority. One of the common challenges for most telcos today is the presence of large legacy estates across these systems, often multiple silos from multiple vendors, a lot of them across multiple generations, and so simplification of these legacy stacks remain a fundamental expectation, when operators start planning for transformation of these systems. And then you add AI into it, which is probably the most recent trend that is actually influencing this particular segment. core commerce is probably the most popular landing spot for early stage innovative examples of AI, be it generative in the early days and and more recently agentic related. So we expect lot more and continued interest from operators worldwide in transformation of their digital touch points.
Robert Curran: yeah, interesting to hear your your thoughts on that. so it's definitely the right time to introduce Hitesh. Hitesh some people know IOH some people may not but it's it's a sizable organization. there's around a hundred million subscribers. eighty-five percent of those use mobile is the primary connection. over nine in ten have smartphones. In your job as the as the SVP of consumer technology and innovation, how do those kind of statistics shape your priorities
Hitesh Kumar: I think more than what is Indosat Ooredoo Hutchison today, we also have to go back a little bit. That is a merger of Indosat, which is an iconic brand of Indonesia, along with Hutch. And what's important was when it came, you get two brands coming into picture, operating into... mobile, prepaid, postpaid, fixed line, FTTH and all. and with different systems, the company is running from decades and all, Hutch was also running for more than a decade, and Indosat is almost more than five decades, six decades, it's been running in Indonesia. One of the considerations which is there for us is to make our systems in such a way that we just follow one philosophy of giving the maximum customer love to our customers. And that means that making everything simple. That's number one. If you want to give great customer love to your end customers, you have to start it with your insight as an organization. Love your employees. Love your customer care people. Love your people who are doing product and marketing and technology specifically sits at a point where you can make things really simpler. Like in terms of friction, in terms of journeys, in terms of the work you have to do to get something out. for a feature functionality. So technology sits there. And with AI coming in last two, two and a half years, it does not just for a telcos like us, it's not shifting us from just delivering what the business asks. It's also shifting us in terms of solving every problem in record speed. It's not about doing things for six months, eight months, nine months. It's about solving it fast, solving it quick. and making the impact count is what counts into our consideration.
Robert Curran: Can you just explain a a bit of the before picture? What did that look like? what did your BSS estate look like? And how did that complexity get in the way of some of the things you're talking about here?
Hitesh Kumar: you know, on Indosat side, had somewhere around 60 million active customers, 60, 70 million active customers. On Hutch side, we have somewhere around 40 million on that case. I imagine these companies running for 10 years, operator running for 10 years, 12 years, 13 years, the system, the processes, the way people work. And the same thing, Indosat was running for more than five decades, same different processes, different systems. we were at a point that we had everything in a SKU of two or three. So suppose we had four or five CRM's for different different things, some legacy, some new, some upgraded, but not just different monetization engine, we had fragmented systems for fragmented functionalities also. we had three billing systems, two charging systems, two mediations, three or four provisioning systems, and all of them were doing different things We had almost five or six order management, eight or nine catalogs. So every... person, every department, every process had their systems, grown with the people, with the processes and matured to a point that, these processes cannot be changed. So that was a architectural spaghetti in terms of giving the same service, giving the same process, but with a different flavor to our customers. So we have two brands, IM3 and 3ID. 3ID is for Hutch IM3 is for Indosat. Now everybody knows Indosat Ooredoo Hutchison is one but if an IM3 customer wanted to go for a service of 3ID SIM card in an IM3 store, we were not able to provide services.
Robert Curran: Mm, mm, yeah.
Hitesh Kumar: If I want to do a SIM swap, for a 3ID card in an IM3 store. We were not able to do that. Why? The systems were different, processes were different. So that kind of sales and distribution was entirely different.
Robert Curran: well let's let's move to the transformation itself. you've been the first to deploy the full stack digital monetization platform DMP from Ericsson. what are the functions that that covers?
Hitesh Kumar: what we have deployed is the world's first full stack for monetization. And it is the world's biggest reference, working reference of this full stack, which is DMP. right now we are running all our 100 million customers, all our IM3, 3ID prepaid, postpaid fix, wireless. on the same platform. The stack is a combination of charging, billing, mediation, order management, and catalog together in one single place. Along with that, we using some of the other components for voucher server, which is for physical vouchers, and some other systems, small, small systems to get this type up and running. What's important to highlight here is that When we went into this journey, we had few customizations in the stack, but when we consolidated our Hutch subscribers, the 40 million onto this to make it 100 million up and running live reference for DMP, we made sure that we used all the standard components of Ericsson stack and removed all the customization as part of it.
Robert Curran: Okay, okay.
Hitesh Kumar: we are fairly happy that we are right now on a stack which is fully non-customized, compliant
Robert Curran: okay.
Hitesh Kumar: with TM Forum Open Digital Architecture. And it's a full stack. It's not like one component from one partner, another component from another partner. Since it's part of the full stack, fully integrated stack, it becomes quite easy for us to scale newer functionalities as well as get agility
Robert Curran: delivered.
Hitesh Kumar: for our business so that we can move faster to the market.
Robert Curran: Yeah, yeah, yeah. Impressive. I know that one of the most delicate parts in this process is migrating because in your case you're not just migrating from one system, you're migrating from multiple systems. and the numbers are big. you mentioned a hundred million or so customers all migrated onto the same platform now. tell us a bit about that. process. How long did that take and how did you manage to to make that a as seamless as possible?
Hitesh Kumar: it was a great achievement with respect to both IOH and Ericsson and also our fellow partner Ecosystem, which we're working in as a part of this whole initiative. it's like when you're doing a heart surgery, you need some of the other specialists also on standby or helping you out in case something goes wrong with some other organ, the specialist has to be there. So. a transformation at this scale are not easy. And typically if you see the industry standards, they are two, two and a half year, three year going into that part. I can simply say that for the prepaid subscribers for 83 million, it was a project of almost, eight to nine months with all the business processes and everything covered. Along with that, migration for 83 million, 83 million I'm talking about installed and active both, 60 million were somewhere active. The migration time was, 18 days. So we migrated all the 83 million customers in 18 days for just for Indosat. Coming back to Hutch, the program of Hutch was much bigger in terms of we were running 18 or 19 parallel programs before. consolidating a core commerce platform from sales and distribution, CX and all. It took somewhere around 13 months or so. And the migration was almost again, 14 to 15 days where we moved all the 43 million migrated on the new platform. What's important is that it is not just executing and tick marking the migration during the program. Indosat took some very strategic calls of simplifying the whole processes, ecosystem, making your brand processes as one and reusing as much as we can. So Blueprint from Indosat the IM3 DMP moving to 3ID DMP and reusing as much as we can helped us to deliver these things in the time we think that are quite industry fastest timeline. at this scale. And I'm very glad and thankful to the team. Saurabh is here. We were both curating the project that our impact to our customers during such transformation was nil. like 3ID migration, has just been one and a half months old, 40 million migration. We got 1,500 subscriber complaints.
Robert Curran: Mm, mm-mm.
Hitesh Kumar: So for 40 million migration, you change the system, you move them and 30 systems were impacted. No mass outages. We didn't have to do any reversals at all. was fairly smooth. So it's the predictability of the delivery and the predictability comes with simplifying as well as being absolutely crisp in terms of what has, what not to do. what we obviously know what to do, but sometimes we don't know what not to do.
Robert Curran: Sure. Mm-hmm.
Hitesh Kumar: And being a very clear, having a very clear hard boundary of what not to do helped us to deliver the transformation in such a short time.
Robert Curran: Yeah, certainly a very impressive and as you say, minimal customer impact. something I picked up on in some of the conversations, was the your use of of GPUs in accelerating the migration process. where did that come from and what was the what was the impact of doing that?
Hitesh Kumar: So I think more than the GPU, Indosat Ooredoo Hutchison in Indonesia is not just a telecom player. We have an affiliate company, Zankore, which is operating NeoCloud, which is a GPU data center. Now we are catering the AI workload demand in the region. for lot of our customers, like for Indonesian customer, for non-Indonesian customer, because you know, AI and the need of AI workload is becoming more and more there. And if you see now data centers are becoming the fundamental layer of infrastructure for your nation's intelligence, and Indosat is pioneering it right now. Now, having said that, there are a lot of AI related use cases, work we are doing. for Indonesia, for IOH, and also for shaping the nation. We have three strategic pillars into our overall strategy. One is AI native telco. We are also an AI native tech co, and then AI nation shaper. So in AI native telco, we do everything with respect to getting our AI embedded deeply in our operating model. We just don't do it with respect to doing use cases. We have a direct P&L impact and we release our results of AI native telco in our shareholder meetings in all our market days and all. The AI native tech co is all about getting AI products into the nation. we have our own large language model. We have. our own app for AI inferencing. So how about AI, we call it, that's the brand. Then also we are doing a lot of products development on top of, which is needed like AI learning platform. There other products which are also getting developed. We launched our token factory for Indonesia, and then the AI nation shaper, which is mainly making sure that Indonesia is the forefront of adopting AI. For that, we are working with the regulators and all here. So when we were doing this migration, we came with a very typical time of like, it's going to take 16, 17 hours to do the data migration. And that's the downtime you have to take. So what we thought was, I think with Ericsson and us, we thought that, okay, the migration is directly, a factor of compute. Why don't we use some powerful compute when we have the GPU? And that's when we started that, let's start processing our migration workload on the GPUs. We have them powered on in Indonesia. And when the team started exploring that thing, I think the results were amazing. From 15, 16 hours of migration. going down to one, one and a half hours of migrating a typical in a migration window, one, one and a half hours, maximum two hours to get that thing transformed and all. sometimes the constraints of resources if removed can really help you to give them business value. we did in three migration windows. If you say that it's 48 hours of downtime. window we have avoided to maybe three to six hours maximum in terms of
Robert Curran: Yeah.
Hitesh Kumar: getting that migration workload. And that's where I think both Ericsson and Indosat we came together. Okay, let's use the technology. Let's be at the forefront not at the back end of it.
Robert Curran: Yep. Very, very interesting. Very interesting. Saurab, welcome to the podcast first. thanks for joining us today. from a technology partner standpoint, what makes the sort of deployment that Hitesh talked about for digital monetization platform different compared to a traditional multi vendor BSS integration migration project.
Saurabh Sharma: So Robert, mean typically such large scale transformation which has so many vendors, so many priorities, so many ecosystems. I think that the real risk sits between the products and the integrations around those products towards the ecosystem be it the northbound or the southbound. So what as Ericsson, we have learnt it as well over the past so many years of implementation. We have tried to separate the core from the non-core. So, DMP as a platform is to a great extent pre-integrated between the core systems before even it lands to the operator. Be it charging, provisioning, order decomposition, the catalog and billing. So, these are pre-integrated to a great extent, which means we are not going to spend time integrating the core when we are on the customer side. the risk of integration is towards the edge, which is only towards the channels or towards very specific southbound or let's say localized integrations, which are different to different operators. This also ties up very strongly towards the concept of minimal customizations and try to stick to the out of the box. And that's where you get the advantage of predictability for scaling of 100 million subscribers. If your core is getting invented on the customer side, the predictability itself will get lost. What we're talking about here for Indosat, mean, we're talking about huge volumes, we're talking about 5 million orders a day, or we're talking about more than 3-4 billion CDRs a day. To get that kind of predictability, your core has to be pre-integrated and that's what Ericsson DMP offers as a start.
Robert Curran: And and just to to clarify something in terms of the
Saurabh Sharma: Yeah.
Robert Curran: operational mechanism today, how is this working? Is this something IOH is running themselves or are you running it as Ericsson on their behalf,
Saurabh Sharma: The operational model is defined based on the domain, the accountability is very clear. So, the fundamental principle is whoever has built it operates it. Whoever has the core competence to operate it should operate it. So, the way Indosat has divided the operation is based by the domains. So, the accountability is very clear. So, Ericsson is responsible for the system that we have implemented which is DMP, includes the charging, provisioning, order decomposition, technical catalog, billing, so on and so forth. And the other partners are responsible for their respective applications. So as Hitesh mentioned, there is an ecosystem that exists. There is a channel layer, there's a middleware, then there are network elements, and each partner is responsible for their respective applications. Now when it comes to Ericsson specifically, we have service operations L1, L2, L3, and L4. And we
John Abraham: great.
Saurabh Sharma: are governed by strict SLAs, KPIs, lifecycle management, performance capacity management. So eventually what IOH gets is that they focus on their core, which is the customer, the product and the growth and the partners who have built the platform end up running what they have built. So they're fully accountable to make it operational, not just the build part.
John Abraham: thank you. Hitesh, thank you for sharing all those bits earlier. Really impressive, especially the numbers that you have shared. I have one quick question for you. you mentioned earlier on that you tried as much as possible to keep things standard, you know, rather than introduce a lot of customizations, which we all know is something that's really, really tough to do, especially in a Environment with a lot of legacy systems, such as the one you started off from. how exactly did you approach it? Because it often involves implications for teams, you know, which is often a big stumbling block or can be, and also process flows and so on. So how big a challenge was that, Hitesh?
Hitesh Kumar: I think it's one of the most subtle thing to do in these kind of transformation, but it's one of the most difficult thing also to do. Like it's subtle, but extremely difficult. And the only way to solve these things is like, see we at Indosat, we believe in one thing, like simplification has to be inside- driven. We cannot put any partner at the spot: that "come and simplify for me. and who drives simplification? To be very frank, it's not a policy. The people running the business or people running that process drive that simplicity, I'll give you an example. When you're trying to choose between two processes from two brands to make it as one process, which process to take? I want to follow an IM3 process or I want to go to a 3ID process. Because the moment you will go from IM3 to 3ID, the 3ID process person or the responsible person will say, my process is better. The IM3 will say, my process is better. So the only thing to handle the subtleness is, I think we follow a simple approach in this. First is data. Where do you have more numbers? Where are you going to get more impact? Where are you going to get more value? That's number one. And data never lies. know, humans are feeling data, have feeling data. Data
Robert Curran: Mm-hmm.
Hitesh Kumar: is black and white. The second thing is communication. we do communication in two ways, We do bottom up communication and we do top to down communication also. So what does that mean? Anything which you decide based on data, you communicate in bottom up. because bottom has to be comfortable. The layer working on the ground has to be comfortable with the new process or with the new change. And that has to be seamless and that has to be simple. And the implications have to be very clear. More than bringing a new technology into the market, it is the absorption which is more important to be planned and executed well. And we have been very clear that nothing top down can work in bottom up if bottom is not. made bottom has not been made comfortable and not being communicated well. And then the top down, top down is important for the sponsorship. If you are not communicating that boss, you are taking a decision to simplify this thing and the implication of doing it and implication of not doing it is not being made clear to the executive classes. The decision will never get the soul. one of the biggest thing which in Indosat was we had, we maintained a very strong executive sponsorship throughout the project. So it was, it was made very clear that if executive senior management executives are not going to decide, we are talking about the chief growth officer. We are talking about the chief marketing officer. We're talking about the chief sales officer, chief operating officer. We're talking about the CFO. We are talking about the CHRO. We're talking about the CEO itself. The CEO used to sponsor the program itself. At the same time, we are talking about the officer at the last mile who has to sell the SIM card today.
Robert Curran: Who has to say this? Okay.
Hitesh Kumar: He's selling from IM3 process, IM3, 3ID process, 3ID. Tomorrow he have to sell everything from IM3. Communicating that, communicating it rigorously and over-communicating to an extent that people get fed up of it is something which, which, which we did relentlessly during the transformation. See, it's like practicing for a year, and your final match is a migration. Because the actual thing hits when you start migrating the traffic to the new system, and you start catering the traffic without any issue. You start catering that traffic coming from those frontliners, from those people who are in the front, their bosses, their bosses. So... That's where the magic happens, you and that is the success mantra for all of us that no compromise in communication. Don't do it in silo. Do it bottom- up, do it top- down and bottom- up/ top- down story has to be exactly same and get everyone together
John Abraham: Yeah. Tip.
Hitesh Kumar: to be part of the problem solving, to be part of the purpose. And this transformation, I think between me, Saurabh and all the people, all the partners who are there. One thing we are very clear, it was 95 % people, I would say, because it needed a lot of processes, changes at every level of it. 5 % and all was more on technology platform and all those things. And that 5 % the moment we carry 95%, I'm very, I'm a little bit on the extra side of people because that 95 % led us. to execute everything on our platform in a very standardized way. And you know, when you ask somebody to kill your own process or or kill your own legacy, the best way is to make it as comfortable as possible. And the best way is to let the people decide that what we are doing is for the greater good of Indosat not killing of process. And that's it, nothing more than that.
John Abraham: Well thanks thanks for that, Hitesh I think it's very impressive that you managed to pull it off
Robert Curran: Now that you are up and running, are you seeing some of the benefits of this simplification and this transformation? What are some of the Examples you can share with us.
Hitesh Kumar: See, a couple of things like our postpaid base grew quite nicely from the time we have moved it to DMP. But at the same time, some of the very strong signs which we are seeing right now is ability to launch products in the market is very quick. in 3ID, we are able to launch a product and we call it as time to market one minute. So the business has that capability. to launch, configure any product, any bundle for any time, any customer, any segment in a minute. It's that easy and that simple that all our business configurations are being done by our business team, not by IT anymore. Compared to if you see any other telcos, it is a very decent cycle of writing a BRD or a IBRS or business requirement document, getting it approved. going to first level of IT analyst, business analyst, then going into the technical analyst, then the system configurators, the testing team and all. All that life cycle of a product launch have vanished. It's like it's not there anymore. The business gets the product construct, designs it, the pricing and does it, gets the approval from business case and all. and configures it right away, it doesn't even arrive to IT. And all supplementary systems of revenue assurance on all are going better. So how does that impact us? I can't disclose all the numbers right now to you Robert, but we are seeing that I was mentioning you previously because of AI system simplification and creating that strong highway. We are reporting our impact on our EBITDA level. exactly because of all these transformations which we are doing
Robert Curran: Saurabh I want to drill into something that came up in in discussion earlier. about the solution that you have. The the
Saurabh Sharma: you
Robert Curran: whole kind of concept-to-cash and again something John mentioned about transforming that process and and Hitesh has mentioned it as well. I I my understanding is Ercisson approaches this in a slightly different way more of a kind of top down than bottom up approach when it comes to to service order management and and customer order management. Can you explain why that distinction matters in a solution like this?
Saurabh Sharma: if I start from the top and then I'll finally go to the bottom. So any order that is not fulfilled is just a promise, right? And the real fulfillment happens at the bottom. So Ericsson's philosophy is to start where it happens and you go backward, right? So where does most see the concept to order mostly fails not at the top. it fails at the bottom layers where the order meets the network because the order has to go through several activation processes with network elements, with regulatory systems, so on and so forth and has to be orchestrated in a perfect way. And that's where the Ericsson order orchestrator, which is EOC, decomposes, orchestrates, makes sure that each of these orders are modeled properly. and they are integrated with the subsystems. Now with Ericsson order decomposition engine, right, so we have a multi-cluster deployment in Indosat. So what it means is we follow where the money is. So each order comes from different channels, right? So based on the priority, based on the revenue, the priority is provided and we make sure that the technical foundation is so strong and then the commercial agility is just a byproduct. So your technical foundation has to be really solid. And if you really look at it, top down approach, we think that it is good for the demos. But when you have to provision 5 million orders a day on a scale of 100 million subscribers, you have to be very solid at the bottom level.
Robert Curran: I'd like to ask something to to John about that because you you touched on it in the earlier part of our conversation, the distinction between things that are specific to telco and things that are generic, can you talk a bit more about some of the problems that you've seen with trying to apply generic systems into a telco context,
John Abraham: sure. I I think one of the most fundamental challenges is that generic systems are not often prepared to support the level of nuances that a telco process might actually require. just in the concept -to- cash, telcos end up dealing with a lot of complex requirements that might change anytime, you know. So a lot of mid-flight changes can happen. that is the issue of COM to SOM handover, which is, you know, can be quite complex, especially when it is a single COM to multi-SOM, multi-vendor handovers, or or some orders for from enterprises could be like multi-site orders with different hierarchies. So all of these are telcos specific nuances that you would seldom find in a outside scenario, you know, outside of this industry. and and so for that reason, a lot of the general purpose systems are seldom prepared to handle this level of exceptions. And even if they do, they are unable to respond in a timely way. when you look at these general purpose generic cross industry systems, they tend to be cheaper upfront. But it is only a little bit along the way that you actually understand that these systems are actually not in a position to support my scale or my level of complexity. I'll also add another point, and it actually goes back to what Hitesh mentioned earlier, when he talked about the type of legacy estate they had before they embarked on this transformation. when you again look at the long term about how how do I make this ready for AI, oftentimes you will find that these processes have to hop between systems, between multi-vendor systems, multi-generation systems, and that's impossible to manage without human intervention, which pretty much removes any of the benefits that we might anticipate from AI. So that's also a key reason, you know, to have a telco-grade system that can that that you can actually map end to end. because then you are actually making it more open to automation, more open to an agentic workflow.
Robert Curran: I we're looking obviously a little bit into the future agentic is very much a hot topic. I'm also aware that IOH Ericsson announced arrangement with Google Cloud to migrate the entire BSS stack into Google. Hitesh, what's the what's the thinking behind behind that and is that is that part of preparing for a more agentic process how did you arrive at that conclusion and what were the concerns that you've managed to overcome obviously things like billing and charging are such sensitive workloads people generally don't want to change them, let alone move them somewhere else. So just tell us a bit about about that decision, how you arrived at that, what you're doing about
Hitesh Kumar: Yeah, Robert, it's a very carefully deliberated thing which we are charting on. know, the industry is looming with skepticism. Okay, can I put charging on cloud or not? I be cloud native is already there, but can I put it on public cloud or not? Is something which is, what happens to the latency, the data, the governance, the... everything comes out of it. Why we want to go on that journey and we want to be evaluating it and we have seen some good results, some certification and all with the GCP. So a couple of things in Indonesia, Google data center is in Indonesia with three zones. That's number one. The second thing is, as I was mentioning earlier, as a part of our strategy, AI native telco. Our whole AI platform is being built with Google on vertex.ai. So all the data decisioning, machine learning models, the inferencing of some of the large language models happen from vertex.ai. All our agents, almost 900 agents or 1,000 agents are there in Indosat right now, as of now as we speak. But all our agents are being scaled on vertex.ai. Our experience is that if you do a cloud native deployment as the right deployment with the right principle and all, business case wise, you're going to be positive compared to running on on-prem and all, as well as you get a lot of agility in terms of your core commerce platform sitting next to your AI platform. And then you can do a lot of things. So the moment you play with on-prem and on cloud platform, everything works, but it's not as seamless as it has to be. And if we have to go at a point of putting more focus of in terms of how my agents are going to run my organization in the next four, five years, then my architecture has to be self-managed, self-governed, lower dependencies on any of the physical infrastructure, which I have when I am doing it on my data center. running in on a world-class infrastructure next to my AI platform. And also the GPU layer for Indosat Zankore is also coming in Indonesia. So look at this like you're working your core commerce workload on GCP, your AI engine or agents on GCP, as well as you have your sovereign AI cloud. which is a GPU data center or new cloud in Indonesia itself. We see endless opportunities in that state of existence. And for that, we are getting Google into the picture with the top level sponsorship, we are getting Ericsson with the top level of sponsorship and we ourselves are also committing to it.
Robert Curran: Fascinating. It's great to have such a a forward looking, well connected, executive led, executive sponsored project and and you've clearly got an eye on on the future as well. there are companies who are trying to take that next stage in their development Hitesh what would you be the one the one takeaway?
Hitesh Kumar: I think one of the biggest takeaway is take care of the people and the processes, the process they are managing. That's the first thing. And when simple way to take care of it is it has to be decision. The decisioning has to be data-led and the decisioning has to be bottom up and top down both. So there cannot be a difference in terms of what the top is thinking versus what the bottom is executing. So that's the number one part of it. The second most important part is AI is no longer a good to have kind of scenario. It is must to have. If you're not designing your system to be agentic in a span of one year to two years of time and all, telco industry will face... the same software backlash, you we have been network driven, not software driven. We'll again get into a point that, just do the basics and all. Somebody will else and come and monetize on top of the infrastructure, the data, what you have, and we'll again lose the battle. So if we are able to do that thing, these two things together, a clear cut value driven approach with respect to people, the data, and the right communication across the organization and. establishing the right level of steering, along with being future ready. Future ready is not just upgrading and moving to a newer stack and all. Future ready means reinventing, re-visualizing how you want to do business, re-visualizing how you want to do a particular thing. Now, with Gen.ai and you can build a lot of things together. But whatever you're building today, if your communications are right, if your alignment is OK, if your data is in the right place, think what's going to run in the next five years on this
Robert Curran: Mm-hmm. Yep.
Hitesh Kumar: case. Because anyway, you're going to spend one year, minimum one year to transform your legacy systems. That one year can be one and a half year, two years if the alignment communication is not OK. After one year, you should be at a point that you are ahead of the curve of the industry. You cannot be slower than what's running outside. Your organization has to be faster than what's running on the outside. And that's the main thing. If those things are taken care, I think it's a matter of time you will achieve the output. Output is guaranteed in that way. So what do you say?
Robert Curran: Yeah, great. Great advice. Suarabh?
Saurabh Sharma: I think great insight Hitesh. I think just adding on top from a delivery organization perspective, the way we see an operator is a combination of channel products and use cases supported by an infrastructure. That's what makes an operator. So my advice to any operator going for a core commerce or a big transformation like this would be your objective has to be very clear. your transformation objective and well connected with the business objective, you have to follow where the money is. You don't do a transformation based on emotions. You do based on what brings value. So there are tens of channels, hundreds of use cases, thousands of products. What is getting the money for the operator? That has to be surgically dissected and focused upon. in the transformation and get rid of anything which is not getting money for the operator. And that's where the rationalization, simplification comes into the picture, which is typically led by the operator themselves. But the inputs based on the data has to be provided as a transformation partner. So why and challenge the obvious, right? Why do you need to do this? You cannot do a like to like implementation just because there has been a legacy vendor for the last 10 years doing a certain operation. You don't have to copy it. But if you go with the copy or like to like, I think that's where it takes three years, four years, and most likely the objective is failed and never happens. I think that would be my recommendation, Robert, from a pure execution
Robert Curran: Mm-hmm.
Saurabh Sharma: perspective.
Hitesh Kumar: And I think one more point on what Saurabh mentioned is I think it's absolute gold what he has mentioned right now. There are two sides of any transformation. One is the operator side of it. Another is the partner side of it. We are creating a growth mindset in our ecosystem. anything which has to be done has to be not on emotion, has to be based on data. So constant engineering of what gives value based on your effort, your effort to value ratio It has to be really good. And that changes the whole course of a good transformation project where both sides work equal. I think we are an example of such a successful large-scale transformation - predictable first-time right, giving value to business and also reducing our non-standardized footprint is a crazy testimony of how it can be done.
Robert Curran: John, fantastic. Thank you all for your time and input today. Hitesh, especially a big project and amazing success. We we look forward to seeing more of your results as you undoubtedly achieve them. That's great. Saurabh, thank you as well for your time today. John, always good to hear from you. very good. Thank you everybody.
Saurabh Sharma: Thank you, Robert.
Robert Curran: that's been fantastic.
Hitesh Kumar: Thank you.
John Abraham: Thanks. Thanks everyone. Cheers.
Podcast Intro 6: you
raspberrymusic tech logo reveal short hi tech intro 414307: you
raspberrymusic tech logo reveal short hi tech intro 414307: you
Robert Curran: Hello and welcome to Talking Transformation, the Appledore Research Podcast. My name is Robert Curran, consulting analyst with Appledor. Today we're going inside one of the most sensitive yet critical programs that any telco can make, transforming its core commerce systems. So joining me today to unpack how operators are navigating this challenge. I have John Abraham, principal analyst, Appledore Research.
John Abraham: Hey Robert.
Robert Curran: Hitesh Kumar, SVP for consumer technology and digital innovation at Indosat Ooredoo Hutchison.
Hitesh Kumar: Hey.
Robert Curran: And Saurabh Sharma, I C T program Director for IOH for Ericsson.
Saurabh Sharma: Hello Robert.
Robert Curran: Welcome, welcome everybody. John, You recently published some extensive research on telco commerce and B2C systems. What's driving this sudden executive urgency?
John Abraham: what we have seen is especially over the past six, seven years, and particularly post-COVID, there's a lot of focus on transformation of the customer interaction model, especially via digital channels, and to the extent that this has become a boardroom priority. One of the common challenges for most telcos today is the presence of large legacy estates across these systems, often multiple silos from multiple vendors, a lot of them across multiple generations, and so simplification of these legacy stacks remain a fundamental expectation, when operators start planning for transformation of these systems. And then you add AI into it, which is probably the most recent trend that is actually influencing this particular segment. core commerce is probably the most popular landing spot for early stage innovative examples of AI, be it generative in the early days and and more recently agentic related. So we expect lot more and continued interest from operators worldwide in transformation of their digital touch points.
Robert Curran: yeah, interesting to hear your your thoughts on that. so it's definitely the right time to introduce Hitesh. Hitesh some people know IOH some people may not but it's it's a sizable organization. there's around a hundred million subscribers. eighty-five percent of those use mobile is the primary connection. over nine in ten have smartphones. In your job as the as the SVP of consumer technology and innovation, how do those kind of statistics shape your priorities
Hitesh Kumar: I think more than what is Indosat Ooredoo Hutchison today, we also have to go back a little bit. That is a merger of Indosat, which is an iconic brand of Indonesia, along with Hutch. And what's important was when it came, you get two brands coming into picture, operating into... mobile, prepaid, postpaid, fixed line, FTTH and all. and with different systems, the company is running from decades and all, Hutch was also running for more than a decade, and Indosat is almost more than five decades, six decades, it's been running in Indonesia. One of the considerations which is there for us is to make our systems in such a way that we just follow one philosophy of giving the maximum customer love to our customers. And that means that making everything simple. That's number one. If you want to give great customer love to your end customers, you have to start it with your insight as an organization. Love your employees. Love your customer care people. Love your people who are doing product and marketing and technology specifically sits at a point where you can make things really simpler. Like in terms of friction, in terms of journeys, in terms of the work you have to do to get something out. for a feature functionality. So technology sits there. And with AI coming in last two, two and a half years, it does not just for a telcos like us, it's not shifting us from just delivering what the business asks. It's also shifting us in terms of solving every problem in record speed. It's not about doing things for six months, eight months, nine months. It's about solving it fast, solving it quick. and making the impact count is what counts into our consideration.
Robert Curran: Can you just explain a a bit of the before picture? What did that look like? what did your BSS estate look like? And how did that complexity get in the way of some of the things you're talking about here?
Hitesh Kumar: you know, on Indosat side, had somewhere around 60 million active customers, 60, 70 million active customers. On Hutch side, we have somewhere around 40 million on that case. I imagine these companies running for 10 years, operator running for 10 years, 12 years, 13 years, the system, the processes, the way people work. And the same thing, Indosat was running for more than five decades, same different processes, different systems. we were at a point that we had everything in a SKU of two or three. So suppose we had four or five CRM's for different different things, some legacy, some new, some upgraded, but not just different monetization engine, we had fragmented systems for fragmented functionalities also. we had three billing systems, two charging systems, two mediations, three or four provisioning systems, and all of them were doing different things We had almost five or six order management, eight or nine catalogs. So every... person, every department, every process had their systems, grown with the people, with the processes and matured to a point that, these processes cannot be changed. So that was a architectural spaghetti in terms of giving the same service, giving the same process, but with a different flavor to our customers. So we have two brands, IM3 and 3ID. 3ID is for Hutch IM3 is for Indosat. Now everybody knows Indosat Ooredoo Hutchison is one but if an IM3 customer wanted to go for a service of 3ID SIM card in an IM3 store, we were not able to provide services.
Robert Curran: Mm, mm, yeah.
Hitesh Kumar: If I want to do a SIM swap, for a 3ID card in an IM3 store. We were not able to do that. Why? The systems were different, processes were different. So that kind of sales and distribution was entirely different.
Robert Curran: well let's let's move to the transformation itself. you've been the first to deploy the full stack digital monetization platform DMP from Ericsson. what are the functions that that covers?
Hitesh Kumar: what we have deployed is the world's first full stack for monetization. And it is the world's biggest reference, working reference of this full stack, which is DMP. right now we are running all our 100 million customers, all our IM3, 3ID prepaid, postpaid fix, wireless. on the same platform. The stack is a combination of charging, billing, mediation, order management, and catalog together in one single place. Along with that, we using some of the other components for voucher server, which is for physical vouchers, and some other systems, small, small systems to get this type up and running. What's important to highlight here is that When we went into this journey, we had few customizations in the stack, but when we consolidated our Hutch subscribers, the 40 million onto this to make it 100 million up and running live reference for DMP, we made sure that we used all the standard components of Ericsson stack and removed all the customization as part of it.
Robert Curran: Okay, okay.
Hitesh Kumar: we are fairly happy that we are right now on a stack which is fully non-customized, compliant
Robert Curran: okay.
Hitesh Kumar: with TM Forum Open Digital Architecture. And it's a full stack. It's not like one component from one partner, another component from another partner. Since it's part of the full stack, fully integrated stack, it becomes quite easy for us to scale newer functionalities as well as get agility
Robert Curran: delivered.
Hitesh Kumar: for our business so that we can move faster to the market.
Robert Curran: Yeah, yeah, yeah. Impressive. I know that one of the most delicate parts in this process is migrating because in your case you're not just migrating from one system, you're migrating from multiple systems. and the numbers are big. you mentioned a hundred million or so customers all migrated onto the same platform now. tell us a bit about that. process. How long did that take and how did you manage to to make that a as seamless as possible?
Hitesh Kumar: it was a great achievement with respect to both IOH and Ericsson and also our fellow partner Ecosystem, which we're working in as a part of this whole initiative. it's like when you're doing a heart surgery, you need some of the other specialists also on standby or helping you out in case something goes wrong with some other organ, the specialist has to be there. So. a transformation at this scale are not easy. And typically if you see the industry standards, they are two, two and a half year, three year going into that part. I can simply say that for the prepaid subscribers for 83 million, it was a project of almost, eight to nine months with all the business processes and everything covered. Along with that, migration for 83 million, 83 million I'm talking about installed and active both, 60 million were somewhere active. The migration time was, 18 days. So we migrated all the 83 million customers in 18 days for just for Indosat. Coming back to Hutch, the program of Hutch was much bigger in terms of we were running 18 or 19 parallel programs before. consolidating a core commerce platform from sales and distribution, CX and all. It took somewhere around 13 months or so. And the migration was almost again, 14 to 15 days where we moved all the 43 million migrated on the new platform. What's important is that it is not just executing and tick marking the migration during the program. Indosat took some very strategic calls of simplifying the whole processes, ecosystem, making your brand processes as one and reusing as much as we can. So Blueprint from Indosat the IM3 DMP moving to 3ID DMP and reusing as much as we can helped us to deliver these things in the time we think that are quite industry fastest timeline. at this scale. And I'm very glad and thankful to the team. Saurabh is here. We were both curating the project that our impact to our customers during such transformation was nil. like 3ID migration, has just been one and a half months old, 40 million migration. We got 1,500 subscriber complaints.
Robert Curran: Mm, mm-mm.
Hitesh Kumar: So for 40 million migration, you change the system, you move them and 30 systems were impacted. No mass outages. We didn't have to do any reversals at all. was fairly smooth. So it's the predictability of the delivery and the predictability comes with simplifying as well as being absolutely crisp in terms of what has, what not to do. what we obviously know what to do, but sometimes we don't know what not to do.
Robert Curran: Sure. Mm-hmm.
Hitesh Kumar: And being a very clear, having a very clear hard boundary of what not to do helped us to deliver the transformation in such a short time.
Robert Curran: Yeah, certainly a very impressive and as you say, minimal customer impact. something I picked up on in some of the conversations, was the your use of of GPUs in accelerating the migration process. where did that come from and what was the what was the impact of doing that?
Hitesh Kumar: So I think more than the GPU, Indosat Ooredoo Hutchison in Indonesia is not just a telecom player. We have an affiliate company, Zankore, which is operating NeoCloud, which is a GPU data center. Now we are catering the AI workload demand in the region. for lot of our customers, like for Indonesian customer, for non-Indonesian customer, because you know, AI and the need of AI workload is becoming more and more there. And if you see now data centers are becoming the fundamental layer of infrastructure for your nation's intelligence, and Indosat is pioneering it right now. Now, having said that, there are a lot of AI related use cases, work we are doing. for Indonesia, for IOH, and also for shaping the nation. We have three strategic pillars into our overall strategy. One is AI native telco. We are also an AI native tech co, and then AI nation shaper. So in AI native telco, we do everything with respect to getting our AI embedded deeply in our operating model. We just don't do it with respect to doing use cases. We have a direct P&L impact and we release our results of AI native telco in our shareholder meetings in all our market days and all. The AI native tech co is all about getting AI products into the nation. we have our own large language model. We have. our own app for AI inferencing. So how about AI, we call it, that's the brand. Then also we are doing a lot of products development on top of, which is needed like AI learning platform. There other products which are also getting developed. We launched our token factory for Indonesia, and then the AI nation shaper, which is mainly making sure that Indonesia is the forefront of adopting AI. For that, we are working with the regulators and all here. So when we were doing this migration, we came with a very typical time of like, it's going to take 16, 17 hours to do the data migration. And that's the downtime you have to take. So what we thought was, I think with Ericsson and us, we thought that, okay, the migration is directly, a factor of compute. Why don't we use some powerful compute when we have the GPU? And that's when we started that, let's start processing our migration workload on the GPUs. We have them powered on in Indonesia. And when the team started exploring that thing, I think the results were amazing. From 15, 16 hours of migration. going down to one, one and a half hours of migrating a typical in a migration window, one, one and a half hours, maximum two hours to get that thing transformed and all. sometimes the constraints of resources if removed can really help you to give them business value. we did in three migration windows. If you say that it's 48 hours of downtime. window we have avoided to maybe three to six hours maximum in terms of
Robert Curran: Yeah.
Hitesh Kumar: getting that migration workload. And that's where I think both Ericsson and Indosat we came together. Okay, let's use the technology. Let's be at the forefront not at the back end of it.
Robert Curran: Yep. Very, very interesting. Very interesting. Saurab, welcome to the podcast first. thanks for joining us today. from a technology partner standpoint, what makes the sort of deployment that Hitesh talked about for digital monetization platform different compared to a traditional multi vendor BSS integration migration project.
Saurabh Sharma: So Robert, mean typically such large scale transformation which has so many vendors, so many priorities, so many ecosystems. I think that the real risk sits between the products and the integrations around those products towards the ecosystem be it the northbound or the southbound. So what as Ericsson, we have learnt it as well over the past so many years of implementation. We have tried to separate the core from the non-core. So, DMP as a platform is to a great extent pre-integrated between the core systems before even it lands to the operator. Be it charging, provisioning, order decomposition, the catalog and billing. So, these are pre-integrated to a great extent, which means we are not going to spend time integrating the core when we are on the customer side. the risk of integration is towards the edge, which is only towards the channels or towards very specific southbound or let's say localized integrations, which are different to different operators. This also ties up very strongly towards the concept of minimal customizations and try to stick to the out of the box. And that's where you get the advantage of predictability for scaling of 100 million subscribers. If your core is getting invented on the customer side, the predictability itself will get lost. What we're talking about here for Indosat, mean, we're talking about huge volumes, we're talking about 5 million orders a day, or we're talking about more than 3-4 billion CDRs a day. To get that kind of predictability, your core has to be pre-integrated and that's what Ericsson DMP offers as a start.
Robert Curran: And and just to to clarify something in terms of the
Saurabh Sharma: Yeah.
Robert Curran: operational mechanism today, how is this working? Is this something IOH is running themselves or are you running it as Ericsson on their behalf,
Saurabh Sharma: The operational model is defined based on the domain, the accountability is very clear. So, the fundamental principle is whoever has built it operates it. Whoever has the core competence to operate it should operate it. So, the way Indosat has divided the operation is based by the domains. So, the accountability is very clear. So, Ericsson is responsible for the system that we have implemented which is DMP, includes the charging, provisioning, order decomposition, technical catalog, billing, so on and so forth. And the other partners are responsible for their respective applications. So as Hitesh mentioned, there is an ecosystem that exists. There is a channel layer, there's a middleware, then there are network elements, and each partner is responsible for their respective applications. Now when it comes to Ericsson specifically, we have service operations L1, L2, L3, and L4. And we
John Abraham: great.
Saurabh Sharma: are governed by strict SLAs, KPIs, lifecycle management, performance capacity management. So eventually what IOH gets is that they focus on their core, which is the customer, the product and the growth and the partners who have built the platform end up running what they have built. So they're fully accountable to make it operational, not just the build part.
John Abraham: thank you. Hitesh, thank you for sharing all those bits earlier. Really impressive, especially the numbers that you have shared. I have one quick question for you. you mentioned earlier on that you tried as much as possible to keep things standard, you know, rather than introduce a lot of customizations, which we all know is something that's really, really tough to do, especially in a Environment with a lot of legacy systems, such as the one you started off from. how exactly did you approach it? Because it often involves implications for teams, you know, which is often a big stumbling block or can be, and also process flows and so on. So how big a challenge was that, Hitesh?
Hitesh Kumar: I think it's one of the most subtle thing to do in these kind of transformation, but it's one of the most difficult thing also to do. Like it's subtle, but extremely difficult. And the only way to solve these things is like, see we at Indosat, we believe in one thing, like simplification has to be inside- driven. We cannot put any partner at the spot: that "come and simplify for me. and who drives simplification? To be very frank, it's not a policy. The people running the business or people running that process drive that simplicity, I'll give you an example. When you're trying to choose between two processes from two brands to make it as one process, which process to take? I want to follow an IM3 process or I want to go to a 3ID process. Because the moment you will go from IM3 to 3ID, the 3ID process person or the responsible person will say, my process is better. The IM3 will say, my process is better. So the only thing to handle the subtleness is, I think we follow a simple approach in this. First is data. Where do you have more numbers? Where are you going to get more impact? Where are you going to get more value? That's number one. And data never lies. know, humans are feeling data, have feeling data. Data
Robert Curran: Mm-hmm.
Hitesh Kumar: is black and white. The second thing is communication. we do communication in two ways, We do bottom up communication and we do top to down communication also. So what does that mean? Anything which you decide based on data, you communicate in bottom up. because bottom has to be comfortable. The layer working on the ground has to be comfortable with the new process or with the new change. And that has to be seamless and that has to be simple. And the implications have to be very clear. More than bringing a new technology into the market, it is the absorption which is more important to be planned and executed well. And we have been very clear that nothing top down can work in bottom up if bottom is not. made bottom has not been made comfortable and not being communicated well. And then the top down, top down is important for the sponsorship. If you are not communicating that boss, you are taking a decision to simplify this thing and the implication of doing it and implication of not doing it is not being made clear to the executive classes. The decision will never get the soul. one of the biggest thing which in Indosat was we had, we maintained a very strong executive sponsorship throughout the project. So it was, it was made very clear that if executive senior management executives are not going to decide, we are talking about the chief growth officer. We are talking about the chief marketing officer. We're talking about the chief sales officer, chief operating officer. We're talking about the CFO. We are talking about the CHRO. We're talking about the CEO itself. The CEO used to sponsor the program itself. At the same time, we are talking about the officer at the last mile who has to sell the SIM card today.
Robert Curran: Who has to say this? Okay.
Hitesh Kumar: He's selling from IM3 process, IM3, 3ID process, 3ID. Tomorrow he have to sell everything from IM3. Communicating that, communicating it rigorously and over-communicating to an extent that people get fed up of it is something which, which, which we did relentlessly during the transformation. See, it's like practicing for a year, and your final match is a migration. Because the actual thing hits when you start migrating the traffic to the new system, and you start catering the traffic without any issue. You start catering that traffic coming from those frontliners, from those people who are in the front, their bosses, their bosses. So... That's where the magic happens, you and that is the success mantra for all of us that no compromise in communication. Don't do it in silo. Do it bottom- up, do it top- down and bottom- up/ top- down story has to be exactly same and get everyone together
John Abraham: Yeah. Tip.
Hitesh Kumar: to be part of the problem solving, to be part of the purpose. And this transformation, I think between me, Saurabh and all the people, all the partners who are there. One thing we are very clear, it was 95 % people, I would say, because it needed a lot of processes, changes at every level of it. 5 % and all was more on technology platform and all those things. And that 5 % the moment we carry 95%, I'm very, I'm a little bit on the extra side of people because that 95 % led us. to execute everything on our platform in a very standardized way. And you know, when you ask somebody to kill your own process or or kill your own legacy, the best way is to make it as comfortable as possible. And the best way is to let the people decide that what we are doing is for the greater good of Indosat not killing of process. And that's it, nothing more than that.
John Abraham: Well thanks thanks for that, Hitesh I think it's very impressive that you managed to pull it off
Robert Curran: Now that you are up and running, are you seeing some of the benefits of this simplification and this transformation? What are some of the Examples you can share with us.
Hitesh Kumar: See, a couple of things like our postpaid base grew quite nicely from the time we have moved it to DMP. But at the same time, some of the very strong signs which we are seeing right now is ability to launch products in the market is very quick. in 3ID, we are able to launch a product and we call it as time to market one minute. So the business has that capability. to launch, configure any product, any bundle for any time, any customer, any segment in a minute. It's that easy and that simple that all our business configurations are being done by our business team, not by IT anymore. Compared to if you see any other telcos, it is a very decent cycle of writing a BRD or a IBRS or business requirement document, getting it approved. going to first level of IT analyst, business analyst, then going into the technical analyst, then the system configurators, the testing team and all. All that life cycle of a product launch have vanished. It's like it's not there anymore. The business gets the product construct, designs it, the pricing and does it, gets the approval from business case and all. and configures it right away, it doesn't even arrive to IT. And all supplementary systems of revenue assurance on all are going better. So how does that impact us? I can't disclose all the numbers right now to you Robert, but we are seeing that I was mentioning you previously because of AI system simplification and creating that strong highway. We are reporting our impact on our EBITDA level. exactly because of all these transformations which we are doing
Robert Curran: Saurabh I want to drill into something that came up in in discussion earlier. about the solution that you have. The the
Saurabh Sharma: you
Robert Curran: whole kind of concept-to-cash and again something John mentioned about transforming that process and and Hitesh has mentioned it as well. I I my understanding is Ercisson approaches this in a slightly different way more of a kind of top down than bottom up approach when it comes to to service order management and and customer order management. Can you explain why that distinction matters in a solution like this?
Saurabh Sharma: if I start from the top and then I'll finally go to the bottom. So any order that is not fulfilled is just a promise, right? And the real fulfillment happens at the bottom. So Ericsson's philosophy is to start where it happens and you go backward, right? So where does most see the concept to order mostly fails not at the top. it fails at the bottom layers where the order meets the network because the order has to go through several activation processes with network elements, with regulatory systems, so on and so forth and has to be orchestrated in a perfect way. And that's where the Ericsson order orchestrator, which is EOC, decomposes, orchestrates, makes sure that each of these orders are modeled properly. and they are integrated with the subsystems. Now with Ericsson order decomposition engine, right, so we have a multi-cluster deployment in Indosat. So what it means is we follow where the money is. So each order comes from different channels, right? So based on the priority, based on the revenue, the priority is provided and we make sure that the technical foundation is so strong and then the commercial agility is just a byproduct. So your technical foundation has to be really solid. And if you really look at it, top down approach, we think that it is good for the demos. But when you have to provision 5 million orders a day on a scale of 100 million subscribers, you have to be very solid at the bottom level.
Robert Curran: I'd like to ask something to to John about that because you you touched on it in the earlier part of our conversation, the distinction between things that are specific to telco and things that are generic, can you talk a bit more about some of the problems that you've seen with trying to apply generic systems into a telco context,
John Abraham: sure. I I think one of the most fundamental challenges is that generic systems are not often prepared to support the level of nuances that a telco process might actually require. just in the concept -to- cash, telcos end up dealing with a lot of complex requirements that might change anytime, you know. So a lot of mid-flight changes can happen. that is the issue of COM to SOM handover, which is, you know, can be quite complex, especially when it is a single COM to multi-SOM, multi-vendor handovers, or or some orders for from enterprises could be like multi-site orders with different hierarchies. So all of these are telcos specific nuances that you would seldom find in a outside scenario, you know, outside of this industry. and and so for that reason, a lot of the general purpose systems are seldom prepared to handle this level of exceptions. And even if they do, they are unable to respond in a timely way. when you look at these general purpose generic cross industry systems, they tend to be cheaper upfront. But it is only a little bit along the way that you actually understand that these systems are actually not in a position to support my scale or my level of complexity. I'll also add another point, and it actually goes back to what Hitesh mentioned earlier, when he talked about the type of legacy estate they had before they embarked on this transformation. when you again look at the long term about how how do I make this ready for AI, oftentimes you will find that these processes have to hop between systems, between multi-vendor systems, multi-generation systems, and that's impossible to manage without human intervention, which pretty much removes any of the benefits that we might anticipate from AI. So that's also a key reason, you know, to have a telco-grade system that can that that you can actually map end to end. because then you are actually making it more open to automation, more open to an agentic workflow.
Robert Curran: I we're looking obviously a little bit into the future agentic is very much a hot topic. I'm also aware that IOH Ericsson announced arrangement with Google Cloud to migrate the entire BSS stack into Google. Hitesh, what's the what's the thinking behind behind that and is that is that part of preparing for a more agentic process how did you arrive at that conclusion and what were the concerns that you've managed to overcome obviously things like billing and charging are such sensitive workloads people generally don't want to change them, let alone move them somewhere else. So just tell us a bit about about that decision, how you arrived at that, what you're doing about
Hitesh Kumar: Yeah, Robert, it's a very carefully deliberated thing which we are charting on. know, the industry is looming with skepticism. Okay, can I put charging on cloud or not? I be cloud native is already there, but can I put it on public cloud or not? Is something which is, what happens to the latency, the data, the governance, the... everything comes out of it. Why we want to go on that journey and we want to be evaluating it and we have seen some good results, some certification and all with the GCP. So a couple of things in Indonesia, Google data center is in Indonesia with three zones. That's number one. The second thing is, as I was mentioning earlier, as a part of our strategy, AI native telco. Our whole AI platform is being built with Google on vertex.ai. So all the data decisioning, machine learning models, the inferencing of some of the large language models happen from vertex.ai. All our agents, almost 900 agents or 1,000 agents are there in Indosat right now, as of now as we speak. But all our agents are being scaled on vertex.ai. Our experience is that if you do a cloud native deployment as the right deployment with the right principle and all, business case wise, you're going to be positive compared to running on on-prem and all, as well as you get a lot of agility in terms of your core commerce platform sitting next to your AI platform. And then you can do a lot of things. So the moment you play with on-prem and on cloud platform, everything works, but it's not as seamless as it has to be. And if we have to go at a point of putting more focus of in terms of how my agents are going to run my organization in the next four, five years, then my architecture has to be self-managed, self-governed, lower dependencies on any of the physical infrastructure, which I have when I am doing it on my data center. running in on a world-class infrastructure next to my AI platform. And also the GPU layer for Indosat Zankore is also coming in Indonesia. So look at this like you're working your core commerce workload on GCP, your AI engine or agents on GCP, as well as you have your sovereign AI cloud. which is a GPU data center or new cloud in Indonesia itself. We see endless opportunities in that state of existence. And for that, we are getting Google into the picture with the top level sponsorship, we are getting Ericsson with the top level of sponsorship and we ourselves are also committing to it.
Robert Curran: Fascinating. It's great to have such a a forward looking, well connected, executive led, executive sponsored project and and you've clearly got an eye on on the future as well. there are companies who are trying to take that next stage in their development Hitesh what would you be the one the one takeaway?
Hitesh Kumar: I think one of the biggest takeaway is take care of the people and the processes, the process they are managing. That's the first thing. And when simple way to take care of it is it has to be decision. The decisioning has to be data-led and the decisioning has to be bottom up and top down both. So there cannot be a difference in terms of what the top is thinking versus what the bottom is executing. So that's the number one part of it. The second most important part is AI is no longer a good to have kind of scenario. It is must to have. If you're not designing your system to be agentic in a span of one year to two years of time and all, telco industry will face... the same software backlash, you we have been network driven, not software driven. We'll again get into a point that, just do the basics and all. Somebody will else and come and monetize on top of the infrastructure, the data, what you have, and we'll again lose the battle. So if we are able to do that thing, these two things together, a clear cut value driven approach with respect to people, the data, and the right communication across the organization and. establishing the right level of steering, along with being future ready. Future ready is not just upgrading and moving to a newer stack and all. Future ready means reinventing, re-visualizing how you want to do business, re-visualizing how you want to do a particular thing. Now, with Gen.ai and you can build a lot of things together. But whatever you're building today, if your communications are right, if your alignment is OK, if your data is in the right place, think what's going to run in the next five years on this
Robert Curran: Mm-hmm. Yep.
Hitesh Kumar: case. Because anyway, you're going to spend one year, minimum one year to transform your legacy systems. That one year can be one and a half year, two years if the alignment communication is not OK. After one year, you should be at a point that you are ahead of the curve of the industry. You cannot be slower than what's running outside. Your organization has to be faster than what's running on the outside. And that's the main thing. If those things are taken care, I think it's a matter of time you will achieve the output. Output is guaranteed in that way. So what do you say?
Robert Curran: Yeah, great. Great advice. Suarabh?
Saurabh Sharma: I think great insight Hitesh. I think just adding on top from a delivery organization perspective, the way we see an operator is a combination of channel products and use cases supported by an infrastructure. That's what makes an operator. So my advice to any operator going for a core commerce or a big transformation like this would be your objective has to be very clear. your transformation objective and well connected with the business objective, you have to follow where the money is. You don't do a transformation based on emotions. You do based on what brings value. So there are tens of channels, hundreds of use cases, thousands of products. What is getting the money for the operator? That has to be surgically dissected and focused upon. in the transformation and get rid of anything which is not getting money for the operator. And that's where the rationalization, simplification comes into the picture, which is typically led by the operator themselves. But the inputs based on the data has to be provided as a transformation partner. So why and challenge the obvious, right? Why do you need to do this? You cannot do a like to like implementation just because there has been a legacy vendor for the last 10 years doing a certain operation. You don't have to copy it. But if you go with the copy or like to like, I think that's where it takes three years, four years, and most likely the objective is failed and never happens. I think that would be my recommendation, Robert, from a pure execution
Robert Curran: Mm-hmm.
Saurabh Sharma: perspective.
Hitesh Kumar: And I think one more point on what Saurabh mentioned is I think it's absolute gold what he has mentioned right now. There are two sides of any transformation. One is the operator side of it. Another is the partner side of it. We are creating a growth mindset in our ecosystem. anything which has to be done has to be not on emotion, has to be based on data. So constant engineering of what gives value based on your effort, your effort to value ratio It has to be really good. And that changes the whole course of a good transformation project where both sides work equal. I think we are an example of such a successful large-scale transformation - predictable first-time right, giving value to business and also reducing our non-standardized footprint is a crazy testimony of how it can be done.
Robert Curran: John, fantastic. Thank you all for your time and input today. Hitesh, especially a big project and amazing success. We we look forward to seeing more of your results as you undoubtedly achieve them. That's great. Saurabh, thank you as well for your time today. John, always good to hear from you. very good. Thank you everybody.
Saurabh Sharma: Thank you, Robert.
Robert Curran: that's been fantastic.
Hitesh Kumar: Thank you.
John Abraham: Thanks. Thanks everyone. Cheers.
Podcast Intro 6: you
raspberrymusic tech logo reveal short hi tech intro 414307: you
