Dr. Devavrat Shah is the Co-founder and CEO of Ikigai Labs and he is a professor and a director of Statistics and Data Science Center at MIT. He co-founded Celect, a predictive analytics platform for retailers, which he sold to Nike. Devavrat holds a Bachelor and PhD in Computer Science from Indian Institute of Technology and Stanford University, respectively.
Ikigai Labs provides an AI-powered platform designed to rework enterprise tabular and time series data into predictive and actionable insights. Utilizing patented Large Graphical Models, the platform enables business users and developers across various industries to reinforce their planning and decision-making processes.
Could you share the story behind the founding of Ikigai Labs? What inspired you to transition from academia to entrepreneurship?
I’ve actually been bouncing between the tutorial and business worlds for just a few years now. I co-founded Ikigai Labs with my former student at MIT, Vinayak Ramesh. Previously, I co-founded an organization called Celect which helped retailers optimize inventory decisions via AI-based demand forecasting. Celect was acquired by Nike in 2019.
What exactly are Large Graphical Models (LGMs), and the way do they differ from the more widely known Large Language Models (LLMs)?
LGMs or Large Graphical Models are probabilistic view of knowledge. They’re in sharp contrast to the “Foundation model”-based AI corresponding to LLM.
The Foundation Models assume that they’ll “learn” all of the relevant “patterns” from a really large corpus of knowledge. And due to this fact, when a brand new snippet of knowledge is presented, it may possibly be extrapolated based on the relevant part from the corpus of knowledge. LLMs have been very effective for unstructured (text, image) data.
LGMs as an alternative discover the suitable “functional patterns” from a big “universe” of such patterns given the snippet of knowledge. The LGMs are designed such that they’ve all relevant “functional patterns” available to them pertinent to structured (tabular, time series) data.
The LGMs are in a position to learn and supply precise prediction and forecasts using very limited data. For instance, they may be utilized to perform highly accurate forecasts of critical, dynamically changing trends or business outcomes.
Could you explain how LGMs are particularly suited to analyzing structured, tabular data, and what benefits they provide over other AI models on this area?
LGMs are designed specifically for modelling structured data (i.e. tabular, time series data). In consequence, they deliver higher accuracy and more reliable predictions.
As well as, LGMs require less data than LLMs and due to this fact have lower compute and storage requirements, driving down costs. This also signifies that organizations can get accurate insights from LGMs even with limited training data.
LGMs also support higher data privacy and security. They train only on an enterprise’s own data – with supplementation from select external data sources (corresponding to weather data and social media data) when needed. There’s never a risk of sensitive data being shared with a public model.
In what varieties of business scenarios do LGMs provide essentially the most value? Could you provide some examples of how they’ve been used to enhance forecasting, planning, or decision-making?
LGMs provide value in any scenario where a company must predict a business end result or anticipate trends to guide their strategy. In other words, they assist across a broad range of use cases.
Imagine a business that sells Halloween costumes and items and is on the lookout for insights to make higher merchandizing decisions. Given their seasonality, they walk a good line: On one hand, the corporate must avoid overstocking and ending up with excess inventory at the top of every season (which suggests unsold goods and wasted CAPEX). At the identical time, additionally they don’t wish to run out of inventory early (which suggests they missed out on sales).
Using LGMs, the business can strike an ideal balance and guide its retail merchandizing efforts. LGMs can answer questions like:
- Which costumes should I stock this season? What number of should we stock of every SKU overall?
- How well will one SKU sell at a selected location?
- How well will this accessory sell with this costume?
- How can we avoid cannibalizing sales in cities where we’ve multiple stores?
- How will latest costumes perform?
How do LGMs assist in scenarios where data is sparse, inconsistent, or rapidly changing?
LGMs leverage AI-based data reconciliation to deliver precise insights even after they’re analyzing small or noisy data sets. Data reconciliation ensures that data is consistent, accurate, and complete. It involves comparing and validating datasets to discover discrepancies, errors, or inconsistencies. By combining the spatial and temporal structure of the info, LGMs enable good predictions with minimal and flawed data. The predictions include uncertainty quantification in addition to interpretation.
How does Ikigai’s mission to democratize AI align with the event of LGMs? How do you see LGMs shaping the longer term of AI in business?
AI is changing the best way we work, and enterprises have to be prepared to AI-enable staff of all kinds. The Ikigai platform offers a straightforward low code/no code experience for business users in addition to a full AI Builder and API experience for data scientists and developers. As well as, we provide free education at our Ikigai Academy so anyone can learn the basics of AI in addition to get trained and authorized on the Ikigai platform.
LGMs can have a huge effect more broadly on businesses seeking to employ AI. Enterprises wish to use genAI to be used cases that require numerical predictive and statistical modelling, corresponding to probabilistic forecasting and scenario planning. But LLMs weren’t built for these use cases, and a lot of organizations think that LLMs are the one type of genAI. So that they try Large Language Models for forecasting and planning purposes, they usually don’t deliver. They offer up and assume genAI just isn’t able to supporting these applications. Once they discover LGMs, they’ll realize they indeed can leverage generative AI to drive higher forecasting and planning and help them make higher business decisions.
Ikigai’s platform integrates LGMs with a human-centric approach through your eXpert-in-the-loop feature. Could you explain how this mixture enhances the accuracy and adoption of AI models in enterprises?
AI needs guardrails, as organizations are naturally wary that the technology will perform accurately and effectively. One in all these guardrails is human oversight, which can assist infuse critical domain expertise and ensure AI models are delivering forecasts and predictions which can be relevant and useful to their business. When organizations can put a human expert in a task monitoring AI, they’re in a position to trust it and confirm its accuracy. This overcomes a significant hurdle to adoption.
What are the important thing technological innovations in Ikigai’s platform that make it stand out from other AI solutions currently available in the marketplace?
Our core LGM technology is the most important differentiator. Ikigai is a pioneer on this space without peer. My co-founder and I invented LGMs during our academic work at MIT. We’re the innovator in large graphical models and the usage of genAI on structured data.
What impact do you envision LGMs having on industries that rely heavily on accurate forecasting and planning, corresponding to retail, supply chain management, and finance?
LGMs will likely be completely transformative because it is specifically designed to be used on tabular, time series data which is the lifeblood of each company. Virtually every organization in every industry depends heavily on structured data evaluation for demand forecasting and business planning to make sound decisions short and long-term – whether those decisions are related to merchandizing, hiring, investing, product development, or other categories. LGMs provide the closest thing to a crystal ball possible for making the very best decisions.
Looking forward, what are the following steps for Ikigai Labs in advancing the capabilities of LGMs? Are there any latest features or developments within the pipeline that you just’re particularly enthusiastic about?
Our existing aiPlan model supports what-if and scenario evaluation. Looking ahead, we’re aiming to further develop it and enable full featured Reinforcement Learning for operations teams. This may enable an ops team to do AI-driven planning in each the short and long run.
