The Center for Data Innovation spoke with Devavrat Shah, co-founder and CEO of Ikigai Labs, an AI platform that helps companies grow and make their business operations more efficient using internal enterprise data. Shah spoke about how Ikigai Labs uses AI to help companies forecast product demand, how the platform helps large enterprises tap into their tabular and times series data, and how current AI trends will impact the company’s trajectory.
Martin Makaryan: What was the inspiration behind Ikigai Labs?
Devavrat Shah: I co-founded Ikigai Labs after spending two decades researching AI, at the Massachusetts Institute of Technology (MIT) where I was particularly interested in graphical model learning and distributed network algorithms. My work there was mostly theoretical, but after working at Netflix, I realized how my academic research could solve real-world problems. I started my first company making AI systems for retailers, but I soon realized that companies across industries could benefit from applying AI to more effectively leverage their data. My experiences building AI products, creating educational programs, and researching AI at MIT all came together when I decided to co-found Ikigai. We wanted to create a tool that would help predict vital business trends, including revenue, sales growth, and time to market. That is how Ikigai was born.
Makaryan: How does Ikigai help enterprises leverage their data?
Shah: Ikigai Labs is a platform that helps enterprises more effectively leverage their data by using AI to extract actionable insights from tabular and time series data. Tabular data is similar to a spreadsheet and with rows and columns. Each row typically represents an individual entry (like a product or person), while each column holds different attributes for that entry (like price, name, or date). Time series data, on the other hand, is a sequence of data points recorded at specific time intervals, which companies use to track changes and observe trends over time. We help companies leverage this data, such as financial reports or supply chain data, using our proprietary large graphic models (LGMs) and large language models (LLMs) to forecast, plan, and help enterprises make data-driven decisions.
Large enterprises have a wealth of data that can help them streamline operations, plan scenarios, and grow their business. But the tabular and times series data that companies keep in various spreadsheets, databases, and cloud data stores are often in disparate areas, which creates obstacles to using the data. The value that Ikigai brings is allowing companies to tap into all of their data through our AI platform and derive insights that would otherwise be impossible or extremely difficult to do. Most of our clients are large companies. For example, we can help a large retailer estimate how much demand a new product will have by using AI to analyze past sales and similar products.
Makaryan: What is the technology that powers the Ikigai platform?
Shah: Our platform is powered by LGMs, which are specialized algorithms that use a graph to represent how different variables are related to and dependent on each other. Unlike LLMs, which work in a more straightforward, linear way, LGMs can capture more complex relationships between variables, unlocking more patterns and insights. We also use LLMs, of course, to make sure that we can make it as easy as possible for customers to derive insights, but LLMs are an added layer after LGMs have already identified patterns and relationships that can inform enterprise decisions. Using time series data, LGMs can help forecast how patterns and relationships may change as the data itself changes over time.
Makaryan: What makes Ikigai Labs unique?
Shah: There are several things that set us apart from other companies offering AI tools or platforms to create workflows and organize enterprise data. First, we are focused on building a comprehensive product that enterprises use to make sense of their data, derive insights, and use those insights for strategic decisions. We do not want to offer companies just a simple tool that uses AI, but rather a product that they can integrate into their core business model. Second, we make our AI system explainable in a way that makes sense to people with no technical background. Our clients can easily access information and understand how the platform made certain predictions. For example, we show the historical patterns behind our forecasts so people can understand and validate them. Lastly, a feature called eXpert-in-the-loop uses human insight from enterprise experts to improve how the platform provides insights through reinforcement learning, thus fine-tuning the product for each company’s specific needs.
Makaryan: How will AI trends impact Ikigai Labs’ future?
Shah: I think AI will be transformative for all sectors, similar to how the cloud transformed how we store and manage data. Similarly, AI will be integrated into software and hardware virtually everywhere. We are already seeing these trends in real-time, and these trends will accelerate. AI models, such as LLMs, will power many products, but these are not the end products themselves. I think as we figure out how to leverage AI effectively, we will realize that we need innovative AI products that address the niche problems different sectors and individual companies face. In other words, an AI chatbot cannot and will not be the only solution for everyone and every problem. This is where I see Ikigai Labs heading as well. We have tried to create a product that leverages the times series and tabular data of large enterprises, and as we progress in the age of AI, more companies will realize how much value they can derive from the data they possess in various forms.
