Cross-IndustrySeason 1 | Ep. 435 min
Entrepreneur and CIO Suvajit Basu and Infocepts’ Shashank Garg on why AI and BI are complementary, what changes for the analyst, and how to start with Generative AI.
Suvajit Basu
Entrepreneur, CIO, IT Leader at CPG Firms
Shashank Garg
Co-Founder & CEO, Infocepts
Also on
Since large language models arrived, a familiar claim has been circulating: BI is dead, and Generative AI will replace the dashboard. Shashank Garg opens the conversation with the other side of it, that most organizations still run KPI reporting and monitoring through spreadsheets or static dashboards, and executive dashboards still dominate board meetings.
Suvajit Basu, an IT executive who spent most of his career at the CPG company Goya Foods and ended as its CIO, argues that AI and BI are complementary. BI accounts for what happened, he says, while AI makes it far easier to move on to knowledge, prediction and forecasting, so the data foundation and the basic reports still matter.
The two discuss what changes for the analyst, how BI, predictive AI and Generative AI work best together, the three buckets of technology, the cost of AI, and Basu’s three pieces of advice for leaders adopting it.
Five Things Worth Your 41 Minutes
Basu’s answer is that BI and AI are complementary, not one or the other. People are coming to realize AI makes certain things easier, but you still need the data foundation, the data pipeline and the basic reports, and then Generative AI is very good at drawing more knowledge from reports. He also says BI is historical, like accounting, and that most dashboards are eye candy, such as a funnel chart or a pie chart, that does not make data actionable.
Garg cites a statistic he read that in almost 80 to 90% of organizations KPI reporting and monitoring is still handled through spreadsheets or static dashboards, and executive dashboards still dominate board meetings. Basu describes the maturity path from data pipeline to reports, dashboards, knowledge, and then prediction and forecasting, which he says AI now makes much easier, for example by letting you ask for the elasticity of a product without first assembling a large set of point-of-sale data.
Basu uses an example from his previous job, where thick BI reports went out on Monday morning and a very smart analyst spent two or three days writing a one-page red, yellow and green summary. That is the part he says will be disrupted: AI will augment the analyst as a co-pilot, and the summary can be available automatically on Monday morning. Garg describes an analyst-assist chatbot for a manufacturing client that combines their semantic layer and BI reports with AI, today through text and soon voice.
Basu says ChatGPT and LLMs are not the only AI, since predictive AI and Generative AI are different and complementary, and gives retail planogram generation and optimization as a clear area. Garg says the biggest wins come from looking at all three together, for example identifying a shopper in real time, subject to privacy laws, using their buying history and profile, and using Generative AI to write a product description tailored to them.
Basu describes three buckets: BI tools such as Microsoft Power BI and Tableau with natural-language queries, data science and machine learning platforms, and the data platform layer such as Snowflake and Databricks. Garg notes that BI tools’ AI features are limited to the data modeled on their own platform, so an organization using three tools may end up with three talk-to-data agents, and that cloud data platforms are adding BI and AI features, with the three buckets starting to overlap.
Basu is positive about where AI is going, but says it is hindered by its high cost, which makes innovation harder for smaller players, though costs are coming down. Garg advises clients to check whether the incremental value is worth the cost, to watch vendor lock-in, to think about who owns the models and IP, to keep monitoring privacy and security, and to invest in change management so that operational managers will trust AI and act on it.
Three things. Start with a use case in which there is a definite ROI, such as demand forecasting, customer segmentation or operational optimization. Focus on governance and data quality, because BI reports hold sensitive data. And build a cross-functional team of business leaders, data scientists and the IT data team, with a program to upskill IT and users. Garg adds that AI learns from humans, so a human in the loop is needed to make it better.
I think BI and AI are complementary. They’re not one or the other.
I think the analyst itself is going to get disrupted with AI, it’s going to augment that analyst, it’s going to be the co-pilot with that analyst
So, we have to have the human-in-the-loop to make the AI better. It’s not going to get better by itself.
The Infocepts Point of View
The conversation reframes the question. It is not whether Generative AI replaces BI, but how the data foundation, the semantic layer and the reports you already trust become the base for AI on top. Where AI changes the work is the step between the report and the decision, which is where the analyst’s days used to go.
Shashank Garg’s guidance is to look at BI, predictive AI and Generative AI together, to check that the value is worth the cost, to watch vendor lock-in and IP ownership, and to treat trust and change management as part of the project, because results arrive only when the person on the ground acts on what AI says.
For leaders weighing this, three questions follow: which report could arrive already summarized on Monday morning, which use case has a definite ROI to start with, and who will own governance and data quality as AI reaches your sensitive data?
Our Business Analytics ServicesOur Data & AI Strategy Services
About the guest
Suvajit Basu
Entrepreneur, CIO, IT Leader at CPG Firms
Suvajit Basu is an award-winning digital transformation and IT executive, recently named Enterprise CIO of the Year at the Inspire CIO Orbie Awards and often described as the Chief Inspiration Officer. His industry expertise spans CPG, supply chain, food and beverage, and media and entertainment. He began at NCR Teradata and SGI, founded a company building ERP software for the media business in New York, and spent most of his career at Goya Foods, ultimately as CIO.
Facing this problem?
Talk to our experts about adding Generative AI to your BI estate without losing the foundation underneath.
#1 on Gartner Peer Insights, 3 years running · 97.2% client retention
Bring This to Your Team
A private briefing on AI and BI for your enterprise, covering use cases, governance and adoption, set against your own data estate.
Shashank Garg 00:00
Hello everyone and welcome to the Intelligent Leader podcast. I’m your host Shashank. And today we have with us, Suvajit Basu. And we are going to have an exciting conversation around the role of AI in the world of business intelligence or AI versus BI. And specifically, look at how AI is disrupting the BI world. Suvajit is an award-winning Digital Transformation and IT Executive who combines strategic vision with hands-on expertise to drive exceptional growth for enterprises. Winner of many prestigious awards, he was recently awarded the Enterprise CIO of the Year at the Inspire CIO Orbie Awards. Congratulations Suvajit, often described as the Chief Inspiration Officer. That’s a new one, Suvajit, I had not heard of that one. Suvajit is known for leading teams with inspiration and translating ambitious ideas into scalable and impactful solutions. His industry expertise spans CPG, supply chain, food and beverage, and media and entertainment. Suvajit, once again, welcome to the show. It’s a pleasure to have you here.
Suvajit Basu 01:23
Thank you. Chief Inspiration Officer was given to me by Bob Evans, who is a great communications person and it’s not really my official title, but I found it pretty amusing.
Shashank Garg 01:40
Yeah, of course. Suvajit, just for our audience, would you like to share a little bit more about your professional journey, so our audiences know a little more about you?
Suvajit Basu 01:48
My professional journey is basically in three parts. And if I go back to my college days, you know, I have a bachelor’s in electrical engineering, then computer science because I fell in love with computers very, very early and then started fortunately went to NCR Teradata. Teradata is one of those huge, huge hardware-based data companies. Right. And that was about 25 plus years ago. So, I was fortunate to work there for four years and after that another four years at SGI Silicon Graphics, the premier company in 3D visualization, et cetera. But they ultimately came up with a product called Data Mine, which was based on ML algorithms and data mining. After that I went on to found a company. We built ERP software for the media business here in New York. That company was acquired by a company out of Silicon Valley in four years and then about 16 years ago went to work for a large CPG company here in New Jersey called Goya Foods and spent the majority of my career there. So ultimately as their CIO. So, it has been a great journey so far and will see what happens next.
Shashank Garg 03:19
Thank you, thank you for sharing that background and once again great to have you here, Suvajit. Just to get started, I was earlier in the day thinking about how the field of data and business intelligence and AI has evolved over the years. Then I read a stat that said that even today we know that in almost 80, 90% of the organizations, the whole KPI reporting & monitoring is still handled either through spreadsheets or static dashboards. And, you know, executive dashboards sort of continue to dominate board meetings. You know, they give you a pretty good job of giving you a view of what’s happened in the past. However, their focus on the present is very operational. They often don’t tell us what really happened, and very rarely would they tell you what’s going to happen next. That’s what you were talking about earlier, what you were doing 25 years ago. And obviously, by solely focusing on historical data, we are missing the opportunities of the future. Right. And just the context of, you know, is certainly missing. Right. I wanted to sort of open up and let you share your view. How do you think the recent advancements in technology and AI, you know, what’s really happening and how can they solve some of these problems? Just give us a lay of the land in your view.
Suvajit Basu 04:47
Well, you hit the nail on the head. You are a practitioner, and it shows. So that is what I think is the heart of the issue, right. BI is historical, right? And remember, we had different stages of maturity in BI, right? Where we build to get the data pipeline going, then we build reports, then we build dashboards, and then we build ultimately from that we create knowledge, right? And ultimately prediction and forecasting, etc. I think it’s only now with advancements in AI that we are getting to do all that much more easily. Before that, we had to deal with mathematical models and so on to make the forecasts, etc. But it’s only now that these large LLMs or large predictive models give us the ability to just ask, what is the elasticity of my product? Whereas before we’d have to get a massive set of POS data and make that happen.
Suvajit Basu 06:03
And just managing that data itself was a huge challenge. Right? So, this is the opportunity that is unfolding right in front of our eyes with AI. And that’s the thing. So, I think AI and BI are complementary technology, like accounting, BI is historical. We are accounting for what happened and then we are presenting that data. One of the challenges with dashboards is, you know, I think most dashboards are like eye candy. You have that funnel chart or a pie chart and that’s it. And there is some tremendous research done by people like Tufte and so on who has spent a lifetime looking at… Steven Few also has done a lot of work in that area, in how to make data actionable. And just by having a funnel chart or a pie chart doesn’t make that data actionable.
Shashank Garg 07:17
I think you made a lot of good points, and I’ll get to the predictive AI in just a little bit. But just the excitement we had a couple of years ago, 18 months ago, around the large language models and you heard statements like BI is dead. You know, Gen AI is going to replace the dashboards. Everybody’s going just wanted to talk to the data using natural language. And now we finally have the power. You will see things like conversational summaries, contextual insights, role-based personalization, and more pattern detection. And, in a lot of ways, you know, we expected by now, 18 months into it or two years into it, Gen AI would just disrupt BI tools. You would see BI platform companies falling off the charts, and replacements happening, but that hasn’t really happened. Right? And you know, it may be because of various reasons, but I’d like to get your take. Do you know what’s happening there?
Suvajit Basu 08:25
Well, to me it’s very simple. I think BI and AI are complementary. They’re not one or the other. Right? So people are coming to realize that AI is making certain things easier. But you still need the data foundation, you still need that data pipeline, and you still need the basic reports. And then if you want to create more knowledge or cull more knowledge out of reports, Gen AI is very good at that. Knowledge of data. Right? So, they’re complementary. I see them as complementary technology.
Shashank Garg 09:08
That’s a good way of putting it for our listeners who are contemplating or who are hearing pitches from Cool Gen. You know, Silicon Valley, VC-packed firms who are claiming they will retire all your dashboards. That may or may not happen anytime soon. Gen AI can certainly make it easy. At the end of the day, whether it’s… AI relies on a very well-developed semantic layer. As you would agree, Suvajit, and as you said, the data foundation and the business layer are on top of it. So, the BI tools have done a great job of creating, managing, and maintaining that semantic layer. And that’s really the hard part. And once that is developed, obviously you can use Gen AI to create wonders. Suvajit, do you have a view on, there’s a lot of technology available, and there are companies that just purely AutoML, like an AutoML capability. What’s their role in this whole mix in your mind?
Suvajit Basu 10:12
Let me give you a real-life example. So at my previous job, what happened is on Monday morning, all these reports, these BI reports came out and it was like this thick, right? And it went to all the C-level executives and key people. Then what would happen is a very smart analyst would basically look at this data, right? And would create an Excel, Word Excel document that, okay, your sellout level for this is this. These are the good points, these are the bad points, these are the things that you should think of in red, yellow and green. It was just a one-page summary of this thick report, but it took two or three days to come out because the person was going through all the data. That is the part that’s going to get disrupted. I think the analyst itself is going to get disrupted with AI, it’s going to augment that analyst, it’s going to be the co-pilot with that analyst which will come from that data. And people don’t have to wait for three days for that analyst report. It’ll be there on that Monday morning automatically available to you. And that’s the opportunity here.
Shashank Garg 11:31
I like the way you frame this. So there’s a role for everything. There’s certainly disruption, not in the whole cycle, but there is certainly disruption at the, what you’re calling an analyst role. Right? So I remember, we supported a client in the manufacturing space and we’ve been working with them for almost 12 years and we manage the whole data platform and the BI layer and the executive dashboards and all that. Right? And just a few, middle of last year we started working with them and say, can you really talk to the data? And the approach there was we create this analyst assist or, you know, so that where we combine the whole work they had done the foundational work, the semantic layer, the BI reports and then combine AI along with it so that you have an Interactive AI-enabled chatbot for the executives to allow tailored natural language interaction. And it’s done through the text right now, but very soon you can get to voice as they enable the multimodal capabilities in their organization, which they haven’t done yet.
Suvajit Basu 12:46
I like that. If I can just give you another example. So I’m an executive, on Monday morning, I’m coming into my… I’m driving in, my drive is half an hour 40 minutes. And in a multimodal AI environment, I can have the AI engine look at my reports and I can ask the questions interactively and the engine can tell me interactively and I can listen and be prepared for that meeting at 8:00 in the morning or 9:00 in the morning. Just like I’m listening to a podcast. But here is my own personal podcast for my company. What I’m going to do this week, who am I going to talk to, what decisions am I going to make and what am I going to suggest making my operations run much better? So, it’s a custom podcast of my reports that I’m going to listen to on my way, and I love that.
Shashank Garg 13:44
The way you put it. I remember a few years ago, I don’t know if you remember this, when Amazon Echo or the Alexa, sorry, Alexa came out, right? One of the first experiments that people did and AWS promoted, they said hey, you could now talk to your dashboards. And that was so bad, that never worked because it wasn’t intelligent. All it was was, was trying to read verbatim and that never worked. But now that you can have an intelligent language model or an AI model-backed chatbot and that’s what we’re experiencing for this company. It’s night and day. You can actually talk, you can actually talk to the data and get to know the state of your business. It can help you prioritize, it can learn quickly even the order in which you are likely to ask questions. So, all of that can be served the way you would like.
Suvajit Basu 14:39
Shashank, in the last two months I have been spending 8 to 10 hours a day with the different AI engines, right? Whether it’s OpenAI, whether it’s Google’s NotebookLM or Google’s Gemini or what Amazon is doing, etc. All of them have their strengths and weaknesses, but things are moving so rapidly, it’s mind-blowing. You and I have lived through the dot com web era, the social media era. However, this is a whole different ball game. And I encourage companies like yours to kind of build a prototype and show it to your clients. And once they see it, I think the light bulb is going to go and say, Shashank, I want it more. I want it more.
Shashank Garg 15:39
Yeah. And I have experienced that Suvajit all of last year. We saw a lot of initial hesitation, especially with the compliance and cybersecurity teams clamping down on usage. Then we saw limited internal usage. We saw some models being brought up on-prem to avoid some sort of leakage of data. But it has happened with the cloud. We see that happening with… There are still issues there obviously, so you must be careful as a CIO, but I’m starting to see increased pickup. And you’re so right. The moment you do a pilot; people get excited and then is when they start talking about possibilities because you are… It’s very hard in this space with the pace at which innovation is happening, to predict or to even lay out the business use case. It’s like you do something and then you think about more and then you do something and then you think about more. Right? It just goes on like that. Suvajit, would you like to discuss any more examples?
Suvajit Basu 16:37
I just want to say one thing from experience, everybody is going crazy about ChatGPT and LLMs and so on. But that’s not the only AI, right? There is predictive AI and then there is Gen AI. So, there is a difference there. Again, those two are complementary things. So, when we look at examples, I’ve seen retail planograms, you know, generation and optimization is a clear area that this kind of technology can play, that can help. Right? Because there is a lot of contextual information that happens with a planogram that is kind of not used except for maybe the dimensions. Right? So, we are infancy when it comes to that. So, I think retail planogram optimization, contextualization, there is a huge, huge potential there. Imagine the number of companies out there, what they’re trying to do in optimization and so on in AI.
Shashank Garg 17:46
Absolutely. What I’ve realized in our experiences is that the biggest wins are coming where you don’t look at traditional BI in isolation, you don’t look at predictive AI in isolation, you don’t look at Gen AI in isolation, but you look at them all together. So something where we are supporting one of our clients on the digital side and the eComm side and the fact that you can now in real-time identify subject to privacy laws, identify the individual married with their buying history profile, put them in a demographic profile and you know what they are gonna like and not like and then use that too, in real-time using an AI algorithm and then combined with generative AI to create product descriptions in real time. So, when Suvajit goes to the same website, you see a different description because of how your buying pattern is or how they have profiled you versus me, where I’m more likely to buy or more likely to see that product in a better way. So those are the kinds of things that excite me, that you’re not just looking at one of these things in isolation but are combining all these three complementary technologies to really create a huge impact.
Suvajit Basu 19:13
So, taking your concept of the digital twin and augmenting that knowledge, there is a huge opportunity to make things better, making decision making abilities simpler, more focused is where we need to go and therein lies that opportunity. Right? We can do… We can get out of that store by buying more things that apply to me faster because time is the most valuable resource, and AI is now giving me back a few minutes of my time.
Shashank Garg 19:47
Absolutely, Suvajit, thank you so much for sharing all those examples. Suvajit, I wanted to shift the conversation a little bit on sort of the state of technology. Technology has a huge role to play in making all the examples and bringing them to life. Would you like to comment on how you’ve seen the traditional technology sort of shifting or evolving? How do you see the current tech landscape? What’s your view there? What’s enabling the shift?
Suvajit Basu 20:21
I think like I was mentioning to you earlier, they’re basically a few buckets, right? Three buckets. One is the BI tools bucket and we all know that that’s Microsoft Power BI tableau and so on, where they’re coming from and then natural language queries after that. The second is the data science and the ML platform, a machine learning platform where we are looking and going through your data set and automatically inferring some things, right? Whether it’s that decision tree, what questions am I going to ask the customer to sell them this boat or this car, right? Automatically. And then the third is the data platform which is the base data layer. So, in that is Snowflake and Databricks and so on.
Shashank Garg 21:16
I like the way you segmented the tech landscape and just kind of what I’m seeing is at least with the traditional BI tools, I’ve seen all of them come out with their AI and Gen AI releases. So, one clicks advanced analytics, anomaly detection, top influencers, and conversational interaction with data. The only problem is that they would limit themselves to the data that is modelled on their platform. So, as a user, if your organization uses three different tools, then you’ll have three different talk-to-data agents in that approach. I’m excited to see stuff like semantic layer enrichment in almost all of them. I’m excited to see and talk about data features. I’m excited to see some contextual summarization now starting to happen within those technologies. I think everybody should be looking to use those and roll out those features if they haven’t done that. I’m also seeing a lot of excitement in the Data Science and AI platforms. The data IQs of the world or the Alteryx of the world, all of them have invested in AutoML capabilities.
Shashank Garg 22:29
So far at least for the simpler use cases where you can predict and data is clean and good. You can sort of reduce the time that used to be required to do the modelling. I think we should all be looking to use that. What’s an increasing trend now, Suvajit? I don’t know if you’ve noticed the Snowflakes and Databricks and even the AWS, the cloud platforms and all their… They started to insert a little bit of BI and AI in their data platforms themselves. So Snowflake came out with packaged offerings like Streamlit integration to build some quick apps like BI apps or reporting apps. They came out with the Cortex AI database has made a play recently, although very recently, they’re starting to add AI plus BI capabilities. That’s what they’re calling it. We’re seeing Amazon AWS making a push with Amazon Q, so AWS Q. So things are certainly starting to get interesting and I believe we are at the point where in all these three buckets you’re starting to see a lot of overlapping features.
Suvajit Basu 23:40
I actually forgot to tell you one of the biggest use cases that I have been…
Shashank Garg 23:45
Yeah, please.
Suvajit Basu 23:47
Working with in AI and BI. For the past, since 2019, 2018. So that’s about six, seven years now. Cybersecurity.
Shashank Garg 24:00
Oh, okay.
Suvajit Basu 24:01
Huge, huge. SIEMS, which were SIEM.
Shashank Garg 24:07
Yeah. Yeah.
Suvajit Basu 24:07
Is where we send all the log information.
Shashank Garg 24:10
Absolutely.
Suvajit Basu 24:11
The BI tools go in and look at it and then try to predict, oh, here is the anomaly.
Shashank Garg 24:15
Yeah. Patterns.
Suvajit Basu 24:17
The NextGen stuff that we have been using is all based on AI models. And that’s basically predicting here is a lateral movement or here is a bad actor or here is a breach. Right? And the best tools. And this is a BI talk, so I’m not going to get into cybersecurity tools. The best tools are the ones that use AI technology at its core to understand that a breach potentially can happen or has happened and then give you an alert, an intelligent alert. Right. And that is huge. Imagine the value of that, so.
Shashank Garg 25:01
No, no, absolutely, you bring up a good point. It’s going to pick up stuff, maybe false positives, but it’s going to pick up stuff for somebody to look at, so.
Suvajit Basu 25:08
Well, that’s where AI comes in. Right? The rule may give you a false positive. The AI decreases the number of false positives.
Shashank Garg 25:15
Absolutely. Suvajit, just to be the devil’s advocate, right? We spoke a lot about possibilities, use cases and how disruption can happen. You know, any callouts? What are you worried about? What are you worried about in this whole context?
Suvajit Basu 25:32
I’m actually very positive about where AI is going and the more and more, we’re using it, I think it’s all positive, it’s net positive because we as entrepreneurs, see the glass half full. So, I don’t think AI in our lifetime is going to take over the world and is going to start shooting at us like a Terminator. I don’t think that’s going to happen at all. So, I take a much more positive view of it. If you look at the brass tacks, one of the reasons AI is kind of hindered right now is because of the high cost of AI.
Shashank Garg 26:17
Yeah.
Suvajit Basu 26:18
So if you do, compared to BI. Right? If you get into AI right now and you want to create a product that will cost, you a lot of money because it takes a lot of computing and so on.
Shashank Garg 26:33
Absolutely.
Suvajit Basu 26:33
And that is making innovation for the smaller players harder to get into. It will push the innovation onto the bigger players. You look at what China has done in the last few years and there’s a great book I read, I don’t know where I have it right now, by Dr. Kai-Fu Lee and he was talking about this five, six years ago, superpowers in AI. Right. And you look at in the last few years how AI, whether it’s from Google, was a pioneer in this area. Microsoft ChatGPT and now Meta what they’re doing and Grok. It’s just unbelievable the amount of innovation and speed at which this has progressed and how everybody is catching up. But who is that small-time innovator, the little player working in their garage? Okay. Who’s making this incredible innovation? I think AI is costly, much more costlier than anything else right now. So that’s one of the things I wish and I’m starting to see the cost of AI governance center coming down. They’re coming down quite a bit. You know, Amazon with their announcements show that their costs are much less, 1/10 the cost of other AI. So costs are coming down those are good things. But that’s a good thing.
Shashank Garg 28:14
Y Yeah, absolutely. I was going to say when I talk to clients, we often advise them, number one, to look at the cost and see whether the incremental value is worth it. And you know now, with the level of innovation is such that the cost parameters change every three, six months, as you said. Right? Nova is, they’re claiming the models are going to be one-tenth of what exists in the market today and that’s disruptive. So I’m excited to see, that we’re starting to do some experiments there and we are actually very bullish on what you know, AWS put out there in terms of their Nova offering and we’ll see a real disruption. That’s what we are projecting. I think we are also seeing that vendor lock-in is, something that you know, clients should continue to, or enterprises should continue to worry about.
Shashank Garg 29:03
Think about who owns the models, think about who owns the IP because in this whole if you’re going use AI to differentiate then you have to own the IP. Needless to say, the whole privacy and security issues which obviously you know much more about that we have to continuously monitor for. And I think lastly, I think the human element and change management and whether when we have all these models and they’re giving the context and they’re predicting the right stuff, whatever we can do to ensure that people are explainable, and humans can really trust. Because eventually, the change will only happen… The results will only come when the person on the ground, that operational manager, is willing to take action and willing to bet their careers on what AI is telling you. I think that I see it to be an area of effort for all of us as professionals helping enterprises get ahead in their AI journey. Great. Suvajit, anything else you’d like to share or sort of advice for anyone who is considering, contemplating, thinking about moving fast or moving ahead in the use of AI, Gen AI and BI in general? What would you like to say?
Suvajit Basu 30:19
I think there are three things. Number one, start with a use case. Start with a use case in which you can say there is a definite ROI for the business. If you don’t see a use case like demand forecasting, customer segmentation, or operational optimization, you know, wait, talk to the business groups, they will tell you what it is. You know don’t just get into it. Number two, focus on governance and data quality. Don’t take that lightly. That’s a very, very important thing because data privacy, security, accuracy, those things are very important. Trust is very important. You know there were a lot of stories about Samsung putting in some IP out there and it was available to everybody. Microsoft copilot, kind of taking HR information or whatever, making it available to the whole company. So be careful, very careful with data governance because you’re gonna deal with very sensitive data with BI reports and so on. So be very, very careful with that. That’s very important. And number three is building a cross-functional team. They will help you get the business cases aligned between IT and the business. Go to the business leaders, have your data scientist with you, have your IT data team with you, work together to create that cross-functional team and then what you need also is a program to upscale your IT department and also your population of users.
Shashank Garg 32:08
Oh yeah, absolutely everyone. Yeah.
Suvajit Basu 32:11
Yeah. To upskill that and have fun projects around that. So, See, Copilot or Gemini or whatever other tools, you know. So that’s very important. So, three things, use cases, data quality, governance and building that cross-functionality.
Shashank Garg 32:32
That’s a great way to sort of package your advice. I love all those three callouts from my perspective. Suvajit, I’m genuinely excited about what all of this means for all of us and enterprises in the coming few quarters. I’m certainly expecting a lot more automation or assistance given to the analysts as you were talking about earlier. So the AI-first automated information insight workflows can really transform our decisions and speed up their actions. So I’m really looking for AI-enabled automation. I’m looking forward to the full data stack. So no more sort of isolated data engineering. You talked about cross-functional teams starting with a use case. So the more full stack you can think about and not silo your… You know remember how departments were siloed earlier. Right?
Shashank Garg 33:27
And even in some organizations, engineering is separate and analytics is separate and AI is separate. All of that is required. But you have to think about all of this as a subsystem for every use case. And the more we can take that approach, I think the more successful all of us are going to be. I’m also a big fan of worrying about the interaction between AI and humans. And I believe that we’re going to see hyper adoption only when… Because AI learns from humans. Right? Their queries, actions, feedback. So, unless we create that hyper-collaboration, hyper, the closed loop. Right. You’re not going to see the benefits that we expect out of hyper-personalization or democratized decision-making. Right. So that AI… Just keep in mind AI learns from humans. So, we have to have the human-in-the-loop to make the AI better. It’s not going to get better by itself.
Suvajit Basu 34:17
I like that human-in-the-loop, the copilot to make things better.
Shashank Garg 34:22
Thank you so much for coming on the show today. Your insights on AI, the call out for predictive AI versus generative AI and the whole impact on the data in the field of BI were incredibly valuable. Thank you for taking the time to share your experiences and expertise with us. To our listeners, we hope you enjoyed this episode of the Intelligent Leader, with me Shashank. If you liked what you heard, please consider sharing it in your network. Do not forget to hit the subscribe button and you can share this with the #intelligentleader. See you next time.
Insights & Resources
Whitepaper
The latest conversations from The Intelligent Leader.
Retail & CPGRetailers generate more data than ever, yet many struggle to turn it into meaningful business outcomes. This podcast explores the hidden challenges…
Retail & CPG 41:16 Ali Miller, VP of Ads at Instacart, and Shashank explore how data, AI, and collaboration are transforming retail media. Ali shares Instacart’s…
Life Sciences 29:56 In this episode, Kabir Patel and Shashank Garg explore the emergence of Agentic AI—intelligent systems capable of autonomous decision-making—and its transformative impact…
About this episode
Not according to Suvajit Basu. He sees BI and AI as complementary: you still need the data foundation, the data pipeline and the basic reports, and Generative AI is very good at drawing more knowledge out of them.
Shashank Garg says that may or may not happen anytime soon. Generative AI can certainly make dashboards easier to use, but it relies on a well-developed semantic layer, which BI tools have done a great job of creating and maintaining, and that is the hard part.
Basu says the analyst role is the part that gets disrupted. AI will augment the analyst as a co-pilot, so a summary that once took two or three days of manual work can be available automatically on Monday morning.
Basu says ChatGPT and LLMs are not the only AI. There is predictive AI and there is Generative AI, they are different, and they are complementary. Garg says the biggest wins come from combining them with traditional BI.
Basu’s advice is to start with a use case that has a definite ROI, focus on governance and data quality, and build a cross-functional team of business leaders, data scientists and IT, with a program to upskill the IT department and users.
Yes. The full speaker-labelled transcript of the 35-minute conversation is on this page, free to read with no form, with chapters that match the key moments.
The Enterprise AI & Data Transformation Partner. Delivering measurable business impact globally.
#1 on Gartner Peer Insights · 97.2% client retention
