So AlphaFold is kind of the biochemist's version of the ChatGPT moment, where large language models and AI get developed and get put to use in a real-world application. ChatGPT is a large language model. These large language models answer us in a very specific way. They actually get to learn our vocabulary by breaking down the English language, and to its kind of core constituents. In this case, you could consider them the 26 letters of the alphabet. Then it identifies patterns and then puts, stitches these letters together in order to coherently answer the questions and the prompts that we give it. Conversely, you can think about enzyme space in the same way. Enzymes are made up of 20 amino acids and you can make the same correlations between these amino acids to get stable structure. We can use these stable structures that turn out to be stable proteins and enzymes. We can use these stable structures to make better enzymes and even select for properties that would up-classify them as to exozymes. So there's a number of different ways you can make an exozyme out of an enzyme. Typically, we refine what we find in nature. So we can engineer for different traits that we want to see. increased thermostability, increased thermostability, increased turnover number, increased catalytic efficiency. We do this by either selecting for a given trait over time, or what we've been having a lot of success in recently is using AI to identify mutations that we can try and select for these traits. It makes the whole process much more efficient and allows us to make a lot of different enzymes that we couldn't make previously. So biochemistry is interesting because it already has its chat GPT moment that we can point to with Alphafold. Alphafold has already proven that we can go from primary sequence of amino acid and determinant structure. At exozyme, we just have to do that next step of going from structure to function. We've developed a lot of key tools in this area. Cell-free protein expression is one of them. where we can quickly go from structure now where we can quickly go from structure now to function. So we can potentially put together the whole pathway from primary sequence to function and that will be able to unleash a number of different exozyme solutions in the future. So the application of large language models or AI for enzyme engineering is huge. Being able to create de novo structures, structural predictions of these 20 amino acids in order to make stable structures is the first step to making bespoke catalysts. Once you have that three-dimensional structure, you can start teasing out or selecting for certain traits. So beyond primary sequence to structure, you can make that leap to primary sequence to function, which is where we want to be. We actually use the proteins proteins as enzymatic catalysts, which means they do a certain chemical transformation. So being able to design these proteins, design these proteins so that they're stable, and then we can select for that final bridge from structure to function, I think is one of the grand challenges that AI and exozymes could potentially solve.