Shareholders of Andretti Acquisition Corp., a special purpose acquisition company (SPAC), have approved the SPAC’s merger with Zapata AI, a Boston-based start-up looking to capitalize on the generative AI explosion by applying “quantum math” techniques and other innovations to processing of generative AI models.
There was no new update on equity value announced for the combined company, and Andretti did not update a timeline for closing the deal, the when it the merger was first announced last September, the DSPAC indicated then that the deal would likely close in the first quarter of 2024.
The shareholder approval comes a little over two weeks after the Securities and Exchange Commission had declared the Form S-4 for the proposed merger “effective,” clearing the path for the deal to proceed.
Zapata AI CEO Christopher Savoie recently spoke during an analyst and investor event held by Andretti and Zapata AI, saying that his company, based in Boston, is looking to become “what we believe will be the first publicly-traded pure-play industrial generative AI company.”
The company’s aim is to make generative AI a more industrial-grade technology that will be more useful for specific enterprise use cases.
“Industrial generative AI is similar to consumer generative AI tools like ChatGPT that are used to generate text and images, but it's tailored to enterprise use cases, taking the generative models behind these popular tools and applying them to critical industrial-scale applications involving language and other forms of data, like sensor data, for example,” he explained. “Industrial generative AI is for challenging problems specific to an enterprise or an industry where we believe a general purpose ChatGPT would not be as useful or sufficient.”
Regarding the example of sensor data, Savoie discussed some of Zapata AI’s work with Andretti Autosport that began before the SPAC merger was proposed. “An NTT IndyCar is outfitted with many sensors that gather data in real time, including factors that are critical to care performance,” he said. “However, not everything that is important to performance can be measured in real time with sensors all over cars moving around the racetrack at speeds that can exceed to 240 mph. One such example of useful information that cannot be measured by a sensor is the slip angle of a car, which correlates to the pace and severity of tire wear while the car is lapping the track.”
Savoie said Zapata AI can use generative AI technology in combination with quantum algorithms to create “virtual sensors” that can provide such data from real-time, real-world measurements that users otherwise would have to rely on simulations and synthetic data for.
As for the quantum connection, Zapata AI was founded in 2017 as a quantum computing software firm called Zapata Computing before it pivoted to generative AI and renamed itself last year. Back in 2022, Andretti Autosport partnered with then-Zapata Computing to explore quantum computing use cases for auto racing, a data-rich sport in which analytics and high-performance computing have become increasingly important in recent years.
Bill Sandbrook, co-CEO and Chairman of Andretti Acquisition Corp (the SPAC is separate from Andretti Autosport, but former racer Michael Andretti has a hand in both), said during the analyst event that after the SPAC was formed in 2021, it was looking for deals with companies in segments that were first-movers in their segments, had a large total addressable market, an experienced management team, and a cross-over with auto racing, and due to its work with Andretti Autosport, Zapata was a “natural fit.”
Savoie added, “We actually started thinking about generative AI as one of the places where you could apply quantum math back in 2017 when we founded the company, but it was hard to talk about it back then. To be honest, back in 2017 or 2018 before ChatGPT, if you said ‘We're doing generative AI,’ people would say ‘like deep-faking? Is that what you mean?’ Not many people understood.”
Zapata AI’s main products are Zapata AI Prose, a large language model generative AI solution similar in operation to ChatGPT, but customized to an enterprise's industry “and its unique problems,” Savoie said. “We believe pros can help companies speed up time-consuming language tasks like filing for regulatory approvals or patents, selling filling in customs forms, or creating consumer documents or reports. We believe it can also be a user-friendly and accessible way to interface with a company's corpus of information or data. Imagine a chatbot that can help you analyze data and create charts or reports simply by asking.”
The company also has a Zapata AI Sense solution that supports technologies like virtual sensors to provide insights, and that “can handle mathematical models, something that ChatGPT and similar LLM applications cannot do very well,” Savoie said. “These models leverage the statistical advantages that come from quantum math, using quantum physics and quantum information science, which can be useful for a lot of different applications, including creating so-called virtual sensors that infer data for variables that wouldn't otherwise be measurable.”
He added that these solutions can be applied to many industries and enterprise use cases.