Tapestry Q&A at the BloombergNEF Summit

A few weeks ago, I attended the BloombergNEF Summit in New York City. It’s always inspiring to connect with leaders across the energy and climate tech industries who are as committed as I am to pursuing innovative solutions to our environmental crisis.

During my trip, I shared more about Tapestry’s mission to make the electric grid visible, and the challenges and opportunities my team faces as we develop our technology. You can check out the video or read our conversation below (note that the written content has been lightly edited and condensed for clarity):

Tapestry is looking to make the world’s electric grid visible. What benefits would that bring, and how are you applying the moonshot thinking that Alphabet’s innovation lab specializes in to your work? It’s stunning that humankind’s largest machine, the electric grid, is largely unchanged from when we first created it. The demands of society have shifted, but the infrastructure remains the same as its inception. To us, this is a beautiful moonshot opportunity: to envision the technology that’s appropriate for the grid we need today. What would be the right fit for this moment, instead of retrofitting our needs to a legacy structure? We believe this is an information and data challenge, so we’re thinking about a moonshot solution around data availability and improving tools for the whole industry to make higher confidence decisions.

Is this all theoretical, or is your technology already being used? One of the most important themes of our work is to make contact with the real world. Sitting in a lab and hypothesizing has its time and place, but we’ve been working with partners since Tapestry’s inception. We’re working with distribution utilities in the U.S., New Zealand, and South Africa, and we’re partnering with Chile’s national system operator to develop a grid planning tool that looks to solve some of the most acute challenges: how do we get things connected? How do we get new loads online? We’re really excited about creating grid planning technology that we can put in the hands of the system operators themselves, so they can make decisions quickly and more confidently. Our transmission planning system has already been deployed in Chile, and we’ve seen them speed up their process. Our goal as a moonshot was to see if we could 100x the time it takes to run a plan. And we’re on our way, with a 30x speed-up in Chile currently.

What does the future grid look like? When Tapestry does reach the moon, all the folks who work in the energy system—generation, managing infrastructure, demand—will have high-confidence and high-reliability information. We’ll be able to connect things to the network in a completely different way. I think of it similarly to the idea of deciding to get internet in your house, but not having to wait six years for that to happen. The grid of the future is a suite of tools buoying the folks who run and manage energy infrastructure, and the societal benefits of faster deployment, lower-cost energy delivery, and higher reliability. The nirvana state for us is around tools that accelerate innovation and solve the industry “trilemma”: making low-cost, high-reliability, and low-carbon energy a reality.

Your work leans a lot on AI. What are the challenges of deploying AI into existing infrastructure such as the grid, which is antiquated? To transform the industry, we need to think about how we get AI research out of the lab and to folks who make business decisions. That's through partnerships and understanding what the needs are. AI is all predicated on data, and projecting forward information requires good information to start with. I hear sometimes that people have too much data, or not enough data, or they don’t trust their data. One of the big challenges and opportunities is how to more thoughtfully and scalably curate and leverage the information and data we are starting to get. And then, once we have good data, how do we make beautiful technology that can actually be applied? 

The third piece is explainability: Every operator needs to feel confident in the assumptions and underlying information that drives recommendation. How do we ensure that where we are using machine learning and AI to drive predictive insights, that the folks making those decisions have high confidence, and it’s explainable where those recommendations came from? Those are the three opportunities we have: curating, organizing, and making the data useful; taking the research and making it applicable, and ensuring everything we are putting together is explainable and understandable so we can move forward with high confidence.