Creative Enzymes has folded AI-driven computational tools into its enzyme R&D workflow and rolled out an upgraded "AI-Powered Enzyme Services" platform. The platform goes after drug discovery and industrial biotech teams—the ones who have wrestled for years with massive sequence space and lousy screening efficiency. The aim: less time, fewer resources wasted on dead ends.

The company puts it simply: to help R&D teams move beyond brute-force screening and zero in on a smaller, smarter set of enzyme candidates.

Enzyme engineering has always bumped up against a basic fact—the number of possible protein variants is astronomical, and a lab can test only a tiny slice. Most projects barely scratch the surface of what's theoretically out there. Creative Enzymes is out to fix that mismatch.

A tighter loop between computation and experimentation

At the center of the new platform is a "Design–Build–Test–Learn" workflow. In practice, that means computational models are used first to narrow down candidate enzymes before anything reaches the lab. Instead of screening thousands of variants experimentally, researchers start with a much smaller, ranked list generated by machine learning models.

The company highlights three immediate effects of this approach:

l Fewer variants moved into wet-lab testing

l Shorter optimization cycles

l More focused use of experimental resources

None of this removes the lab work—but it changes what gets tested and when.

Discovery, engineering, and design in one system

The platform itself is split into three main modules.

One focuses on AI-Guided Enzyme Discovery, combining sequence database mining with structure prediction tools such as AlphaFold and protein language models (e.g., ESM-2). According to Creative Enzymes, this approach has pushed hit rates for functional enzyme variants above 50% in internal case studies.

Another module is aimed at enzyme engineering. Here, machine learning models—including Gaussian process regression (GPR) and other surrogate modeling techniques—are used to predict how specific mutations might affect performance. In one example shared by the company, work on fructosyl peptide oxidase (FPOX) led to a Tyr261Trp variant. As reported by the company, this mutation increased specific activity by 5.1 times and improved catalytic efficiency (kcat/Km) by 11.7 times under its internal assay conditions.

The third module uses generative AI to design novel enzyme sequences based on existing natural scaffolds. These are intended for reactions lacking a characterized natural enzyme, expanding the scope of enzyme applications in industrial chemistry and synthetic biology.

Still grounded in laboratory validation

Although it focuses on the computational aspect, Creative Enzymes indicates that nothing is finalized in silico. The candidates produced by AI are still expressed, purified, and tested. Kinetic data and stability data, including kcat, Km, Tm, etc., are then incorporated into the models.

Eventually, this establishes a design-and-build cycle of experimentation, where each experiment refines the quality of predictions. Based on internal benchmarking across multiple enzyme engineering projects, the company states that this method can reduce the amount of experimentation by approximately three to five times compared to traditional directed evolution workflows.

Current applications

The platform focuses on three primary sectors: pharmaceutical production, industrial biotechnology, and synthetic biology.

In pharma, the focus is on enzyme-based pathways for API synthesis. In industrial biotech, the focus is on robust biocatalysts for harsh production environments. In synthetic biology, the focus is on constructing biological pathways to synthesize new compounds or enhance yields of existing ones.

About Creative Enzymes

Creative Enzymes provides comprehensive enzyme solutions, powered by AI-Driven Enzyme Engineering Solutions, for research and industrial applications worldwide. The company combines enzymology expertise with computational modeling and automated laboratory workflows, covering the full process from enzyme discovery through to scale-up.