10.25.23-Digital-Zapata-AI

YUDONG CAO: Thanks for the intro and thanks for the organizer for putting together this conference. We're Zapata AI, the industrial generative AI company. We spun out of Harvard in 2017 primarily focusing on quantum computing and the implication of quantum computation.

So that's another very disruptive field, which is adjacent to the topic of generative AI. And over the past years, we have the fortune to work with MIT on several fronts. We're very fortunate to spend our early years at The Engine along with some initial investments. And also, we have worked with different departments at MIT on different collaboration projects.

So briefly summarizing, we are focused on some of the most mathematically complex computational problems that enterprises face, a lot of which involve what we today call generative AI. But starting five years ago, we have already been developing generative modeling techniques, leveraging quantum and quantum physics related techniques.

So I want to tell you more about how we think about our impact in generative AI today. So our broad goal is to aim at these advanced mathematics and computing technology. And our offering is built on two pillars. One is the algorithm, the proprietary, scientifically deep insights that we encode into our libraries and also the computing platform for enabling these computation in various infrastructures.

So we think about our offering in terms of several tiers. So a lot of the generative AI conversation these days is exclusively about language, LLMs and ChatGPT. But our viewpoint starting from five years ago is that generative modeling is far more than just languages. It's about time series. It could be about images, high dimensional data. So we want to really recognize that.

So in addition to PROSE, which is exclusively focused on large language model related work, we also have a vast collection of different algorithms that we call SENSE, which is leverage complex mathematical models for problems such as anomaly detection, image recognition, time series analysis. And then also to reflect our commitment to research on understanding delivering quantum advantage, this is essentially the source of our innovation.

And then to capture all of that in a software platform is what we call Orquestra, so this is our tool and infrastructure for developing and deploying industrial applications. So there's a lot to cover. And it's very difficult to succinctly describe what we have done in the past five years. But let me highlight a few key points.

So with SENSE, our goal is to help enterprises make more informed decisions by enriching business analytics with new intelligently generated data. So in many cases, businesses benefit from being able to answer what-if questions. And a lot of the use for generative modeling and generative AI essentially is for answering "What if we had this input for which we have very sparse data for?"

And we have been developing models in different application areas. One of those is what we call virtual sensors, which means how we can use data that are measurable to infer data that are otherwise very difficult to measure. So I'll give you one example, which is our collaboration with Andretti Autosport. So for those of you who are motor sports fan, you probably have heard of the name Andretti.

So they are one of the top teams in IndyCar and also working their way into Formula One as well. So we have a very deep partnership with them developing generative AI technologies for what we call virtual sensor, essentially for using AI to predict quantities that otherwise would be very difficult to measure. And we have built edge solutions that are deployed on the race analytics command center that is in the racetrack as well.

So we're deeply embedded with the team. And we're very happy to share their success. This has been a partnership for two years. And we very much intend to go forward.

And I'm hoping to tell you a lot more. So this is just a short list of the sectors and industries that we target. And we target a very diverse set of global business.

So essentially, the intuition is that any business that arrives at a certain complexity will involve complex mathematical problems, such as optimization, and more recently, generative AI. And we won't be able to do it without a mature partner ecosystem that is offering us with a variety of computational hardware. So this is part of the challenges that we addressed with our Orquestra platform.

So we started off working with quantum computing, so as you see, the QPU here. But more recently, we've started integrating GPUs and TPUs for these more advanced generative AI capabilities. And also, we have very mature set of collaborations with research with universities around different research projects, with MIT on the BMW collaboration.

So Pat Bernard and I will be and also Tim Hirzel will be at the booth at the lunch. So we welcome any further conversation. Thank you very much.