Private Company Spotlight
Moving from LLMs to LGMs with Ikigai
Needham Industry Update: SaaS and Appliction Software
Last update: April 30, 2024
NOTE: This is an excerpt from the Needham Industry Update for SaaS and Application Software, published on April 30, 2024. The full report can be found here.
Private Company Spotlight: Moving from LLMs to LGMs with Ikigai
In this industry note, we highlight Ikigai, an early-stage private company in the broader General AI software space. We had the opportunity to sit down with co-founder & CEO Devavrat Shah and President Kamal Ahluwalia. Mr. Shah is a professor at MIT teaching statistics, machine learning, and modern parlance AI for the last 20 years. Mr. Ahluwalia created strategy and GTM initiative during the early growth stages of several well-known Silicon Valley software startups over the last 20 years. We believe public and private investors alike should familiarize themselves with companies like Ikigai because these companies are at the forefront of driving differentiated operational value for companies versus the predominately content-driven LLM use cases popping up across most of our coverage universe today.
Ikigai's platform is based on Large Graphical Models (LGMs) versus today's better known Large Language Models (LLMs). Where LLMs are based on language, LGMs are affectionately called LLMs for numbers. LGM platforms like Ikigai leverage structured tabular time-series data that are often mathematical in nature making them computationally efficient. Ikigai then leverages stochastic modeling to present data and predict outcomes that account for uncertain levels of predictability or randomness. A key difference between enterprise use of LLMs and LGMs will be moving from workflow optimization which is a key LLM feature of content creation to the data centric LGM output driving a business to take a certain action.
The output of these LGMs is a generative, probabilistic answer to key business questions across several operational departments and horizontally across industries. Today, Ikigai can yield insights into key scenario questions like supply and demand planning, drug shortage planning, improved financial risk management for fraud or cash management, and even modeling/planning people's skills versus people tasks.
Another key difference to highlight between LLMs and LGMs like Ikigai is the cost and infrastructure requirements. Ikigai targets smaller enterprise data sets enabling Ikigai to run on more cost-effective CPUs and not the large GPUs chasing the LLM content use cases. The primary difference here is Ikigai only needs to access a company's own data and model in the influence of external factors (weather for instance) versus petabytes of historical novels or historical content a LLM typically requires.
Key Points:
- In layman's terms, Ikigai creates Generative AI for structured data
- How does a LGM differ from an LLM?
- Why we believe LGM Gen AI technologies can be more impactful to Enterprise environment than LLM
- Broad Use Cases Drives Significant TAM
- Trusting and explaining the results from a GenAI platform remain key for adoption
- Willingness to spend on AI initiatives remains cautious as companies seek funding, to better understand operational impact, and of course the ability to keep data secure
- Modern vendor competition limited; AI space to expand beyond legacy players with time
Where LLMs are based on language, LGMs are affectionately called LLMs for numbers.
Click here to read the full paper from Needham & Company
About Needham & Company
Needham & Company is a globally recognized investment banking and asset management firm focused solely on growth companies and their investors. Founded in 1985, the firm is headquartered in New York City with offices in Boston, Chicago, Minneapolis, San Francisco, and Miami
Needham’s mission is to provide its clients with the long-term advice they need to achieve their business goals. The Firm is committed to exceptional service and is born of a tradition which stresses integrity above all else. We strive to be front of mind, approachable, and idea driven.