AIP Podcast EP 62 - Instant Actionable Insights for the Supply Chain with Lumi AI

This episode's guest, Ibrahim Ashqar, Co-founder and CEO of Lumi AI, discusses how Lumi's platform makes it easy for data teams to enable conversational self-service analytics. This allows non-technical users to access specific insights without SQL/Python expertise and inaccessible through centralized dashboards, freeing data teams for strategic work. By eliminating manual analytics, mid-to-large enterprise clients using Lumi have seen a 38% reduction in procurement costs annually, discovered multi-million dollar inventory discrepancies, and significantly accelerated data-driven decision-making.

Follow AIP Affiliate, Lumi AI
Website: www.lumi-ai.com
LinkedIn: https://www.linkedin.com/company/lumi-ai/

Follow AI Partnerships
Website: https://www.aipartnershipscorp.com/
LinkedIn: https://www.linkedin.com/company/aipartnershipscorp/
X: https://twitter.com/AIPartnerships

The AIP Podcast is hosted by Anne Cheng, on behalf of the AI Partnerships, a Railtown company

Transcript

SPEAKER_01 0:05

The term business intelligence convers out images of dashboards, turning out insights and more data visualizations that help us keep a pulse on everything. But it's often just that dashboards aid our understanding, they throw out more questions than the answers they provide, leading to this term called dashboard anarchy. A situation when we have more and more dashboards, but less true understanding of what the data is really trying to tell us. Good day, everybody, and welcome back to yet another episode of the AIP podcast. Once again, it's Anne, your host of Supercharged Lab. And on the on behalf of the AI Partnerships Corps, in today's episode, we are going to be meeting Ibrahim Ashkar, who is building Lumi AI. Ibrahim is a data and analytics leader with strong experience building enterprise-grade machine learning-based data products for companies with intensive supply chain operations. As a compelling communicator with a strong ability to lead teams, driving effective results, and convey complex technical concepts to a non-technical audience. Hell we need that. Hey Ibrahim, what a thrill and welcome to the AIP podcast. Tell us your story. What drove you to embark on the road to building your own company, Lumi AI?

SPEAKER_00 1:53

Yeah, for sure. And thanks for having me, Ann. It's a pleasure to be here. You know, prior to founding Lumi, as you mentioned, I spent my entire career working data analytics. Uh, right before this, I was a director of data science at a unicorn supply chain technology startup scale-up called Stored. And before that, spent time in consulting within Deloitte's artificial intelligence practice, uh, building data products for Fortune 500 type clients. And that's actually where I met my uh co-founder as well, Tudor. We were on the same projects delivering uh you know data products to various retailers and brands and transportation companies. And really, we built Lumi to solve a problem that we've seen at everywhere everywhere we've ever worked at or consulted for. Um yeah, and uh you know basically what we noticed is that the term business intelligence is often synonymous with dashboards, but the reality is more dashboards do not mean more insights. And if anything, having too many causes confusion and distrust in the data. And so what we've done with Lumi essentially is the data analytics software that helps teams explore operational data, generate custom reports, and extract insights with just plain language prompts. No SQL or Python needed, really helping folks unlock hidden value, boosting productivity, and reducing reliance and overburdened teams. So the the whole reason why we started Lumi is essentially solved problems uh that I was facing when I was in my roles and tutor as well. So it's uh it's a bit about us.

SPEAKER_01 3:27

That's amazing stuff. Tell us more about Lumi AI and the key challenge that it is undertaking and why this is such an important mission.

SPEAKER_00 3:38

Yeah, for sure. And I I really think I'll go back to that misconception that everyone seems to have around business intelligence and how that's just you know, think of business intelligence, you think of dashboards. Um, but yeah, you know, having too many dashboards just causes confusion and distrust. I've even heard certain C-suite executives refer to it as dashboard anarchy and often refer to their organizations as being data rich but insights poor. Now, don't get me wrong, having centralized dashboards is often a useful starting point, but the real insights, the ones that move the needle, still require hours of spreadsheet analysis and often abilities in coding, SQL or Python, skill sets that's usually reserved by data teams who are just constantly swamped with requests. That's a reality I you know I experienced firsthand when I was at Stored. I was obsessed with this concept of enabling self-service analytics, this utopia where all executives, supply chain teams, and anyone who doesn't know how to code can kind of find valuable, actionable insights hidden in the data on their own. But quickly realized that this utopia was a bit out of reach given the existing tooling that we had, because no matter how many dashboards we cranked out, our team of 12, which is a sizable data function, still kept getting bombarded with customer requests from across organizations. Um, it's a much larger organization, so it's you know a situation where it's a few serving the many. And about 40%, if I recall correctly, of our weekly sprints, I had to allocate and dedicate to ad hoc requests versus strategic planned work. And this figure varies from companies to company. Some some some have entire data teams whose entire role is simply responding to data requests submitted by business teams. And so this isn't just an operational problem, it's really a missed opportunity to uncover tangible value hidden in the raw data.

SPEAKER_01 5:32

That is so important to really find value that is hidden in data and overcoming dashboard anarchy. Let's dive deeper into this. What is dashboard anarchy and really also the real elephant in the room? Aren't all enterprises embarking on their AI journey trying to find value hidden in the data? What sets you apart from all the rest of the AI wannabes?

SPEAKER_00 5:58

Yeah, great questions there. Um, I'll try to adjust them in sequence here. So dashboard anarchy really refers to that situation where business teams have too many dashboards, right? Some are too high level to be valuable, others are too complicated to understand, and the remaining are either outdated or simply just plain wrong, and that happens all the time. And instead of bringing clarity, it creates this maze of complex data sources that just confuse the users rather than empower them with actionable insights. Now, you're you're totally right, all enterprises are embarking on their AI journey to finding value hidden in their data. Um, and if they aren't, they should, uh, because there's lots of value everywhere. And uh, you know, I think you're gonna maybe drill down a bit deeper into the actual use cases. There's a lot kind of going after the the uh insights extraction from unstructured data sources like Google Drive, SharePoint, Notion, and that's great. You know, companies like Glean are super focused at that. At Lumi, we're we've become specialized and quite well versed in extracting insights from large structured data sets stored in ERPs like Oracle, SAP, Microsoft Dynamics, or from data warehouses, depending on the client's maturity, if they've migrated the data over to a data warehouse or not. Now, some organizations have tried to build this in-house, these you know, talk-to-year data capabilities, structured data capabilities, but it's truly quite tricky, and there's a lot involved in building a system that can mimic uh the task that a skilled human analyst can do. And that's really a part of our differentiation, I think, our agentic workflows that have been equipped and they're quite sophisticated now to really do a lot of the things that you'd expect a competent senior analyst to do. But the space is noisy and can largely be grouped into two buckets, I would say. You have your traditional BI tools and you have your emerging data chat bots. The traditional BI tools are now layering in more generative AI capabilities and features, but they're mostly focused on streamlining the process of creating a dashboard and not insights generation, right? And those are two fundamentally different things. You know, analogy here is like do you want a better shovel or do you want actual gold, right? Um, as with regards to the emerging data chatbots, um, we've seen that they seem, you know, there's many out there. Some are simple and more focused on hobbyists and not enterprises. So it's a different entire demographic altogether. Some can't connect to the same systems as us, and some are just focused on developers and not catering to any business users. Uh, what in the case of Lumi, we're truly focused on that insights extraction from these large, complex, and often inaccessible operational databases. The workflows that we've built are sophisticated. We call them agentic workflows because that's the that's what's actually powering the technology, a network of agents, each specialized to complete certain tasks and collectively mimic what a human senior analyst would be capable of doing, but much, much faster. Um, and we've actually benchmarked our performance against other vendors claiming talk to data, and Lumi quantifiably outperforms in direct head-to-head comparisons when we use the same data sets and even ask the same questions. And really, that's primarily a function of our technology that includes the prior agentic workflows and another aspect which we call the knowledge base, which is the repository that stores essential context about data structures, business terminology that allows this network of agents to understand users' intent and be able to accurately answer users' questions. So that's really at the heart of our technology.

SPEAKER_01 9:41

That's amazing. Let's perhaps talk a little bit about the use case now. Um, how have you demonstrated success in working with your clients? Perhaps maybe if you can think of your past experiences within consulting and in data and analytics roles, how has that actually helped you on this journey as well?

SPEAKER_00 10:02

Yeah, without a doubt. I I think it's all about helping them unearth tangible value, right? Time savings uh or cost avoidance, and that's that's fairly useful, right? Productivity boosters. But I really think in order to be able to effectively sell into enterprise, you must be able to hit the bottom line somehow. And that that you know, we're super laser focused on that at Lumi, helping uncover tangible value, surfacing intelligence that can either free up cash flow, reduce costs, and increase revenues. And we have examples across all three there. So for freeing up cash flow, we've helped one of our mid-sized uh CPG clients in a chocolate factory uh free up cash flow by uh identifying excess and obsolete inventory uh where there's like raw materials or packaging materials where the current inventory level just exceed consumption over the past six months and recommending that it should you know not be procuring more of that of those raw materials or packaging because it's just gonna put them in uh a further constraining their cash flow, put them in a further excess or obviously inventory position. Uh, in reducing costs, uh, we've helped a textile, a medium-sized textile manufacturer uh reduce procurement costs by about 38% per year. Uh and in this example, this case study, Lumi actually found pricing discrepancies between vendors who sell the same raw material and recommended either negotiating or consolidating to the vendors offering better prices. So very clear way, very easy way to just reduce procurement costs there. And then finally, on the increased revenues, or rather, in this particular one, minimize loss sales, um, helped a major Fortune 50 retailer uh who was struggling with on-shelf availability issues uh improve uh their in-stock percentage. And by the way, this was an issue that was quantified to impact about 8% of annual revenues and where about 75% of the reasons were caused by store operations. And in an attempt to address this problem, the original state, uh, the account management and retail ops would have to manually export data from multiple dashboards, consolidate, analyze the data, in some cases, bring in someone like an analyst to provide custom data that's not been available from the dashboards, and then finally, after a week, uh generate this list of stores who aren't following inrich protocols properly. And with Lumi, we had streamlined this entire process down to just a few minutes by asking a couple of questions in plain language, significantly boosting productivity and helping recover uh some of the lost sales attributed to these stockouts in the process as well.

SPEAKER_01 12:35

Amazing. Well, we're coming up to the end of our time, so one for the road, what are your plans for Lumi AI?

SPEAKER_00 12:43

For sure. I think I think everything we're working on right now is about releasing features to improve the overall user experience and continue reducing the time to value. And so this also expands into enabling more automated insights and improving Lumi's abilities to responding to more vague, higher order, complex questions. Um, after all, the long-term goal here is to be enabling actions, right? And this is the promise of agentic workflows. We want to build a system that can autonomously extract insights from operational data sets, communicate opportunities and areas of concerns to relevant stakeholders, and execute the strategic business actions based on the recommendation surface of the users. Um it's a long journey, but we're well on our way, and uh, we're super excited. This entire space, there's so much going on, but we uh we're happy where we are and you know, just every day working harder towards that end vision and goal that we're going after.

SPEAKER_01 13:44

That's great. Well, Ibrahim, thank you so much for joining us today. I surely had a lot of fun, and I hope you did too. And to all of our listeners and subscribers on Spotify, YouTube, and LinkedIn, your support has been incredibly valuable to us. And please don't forget you gotta like, share, follow us, and share this with somebody uh someone who would probably benefit from it. It helps us so much more than you'll ever know. So, once again, my name is Anne. I'm your host here from Supercharged Lab, and on behalf of the AI Partnership Support, I happen to speak to Ibrahim Ashikar of Lumini AI. Thank you for sharing your time with us. Thanks, Anne.