Structural screening and functional interpretation
Sequence screening looks for matches to known threats. We've built structural screening to identify similarities that sequence comparisons miss, and are developing model interpretability methods to assess biological function.
Predict the fold, then compare it
We predict how a sequence folds, then compare its structure with known hazardous proteins. Foldguard is the only dedicated structural DNA synthesis screening product. The tool is built and currently in testing.
Find functional signals inside biological AI models
We're developing methods to analyze a model's internal representations for signals of biological function, including functions that sequence or structural comparisons may miss.
Applications in biosecurity and therapeutics
Screen sequences before they leave a model or enter synthesis
We're developing functional screening to flag hazardous activity in generated sequences and synthesis orders before a design reaches the lab.
Guide the design of therapeutic sequences
We're developing methods to guide biological AI models toward therapeutic targets, such as phages active against antibiotic-resistant bacteria and oncolytic viruses.
For AI labs and DNA synthesis providers
Evaluation, red teaming, training, and live screening
We're developing three offerings for AI labs: biological risk evaluations and red teaming, training support, and live screening of generated sequences. These tools are being designed to run within your VPC.
Effective without being overly broad
Broad biosecurity filters can block legitimate research. We're developing function-based screening to make those restrictions more precise.
Designed to fit existing screening and compliance workflows
Our structural screening tool is designed to complement sequence checks. Functional screening is in development.
Built for where screening requirements are heading
Screening frameworks are moving toward shorter sequences and broader definitions of biological risk.
AI is showing biological capabilities we haven't seen before
One commercial biosecurity screening tool flagged less than a quarter of the AI-redesigned toxin sequences tested.
The gap is already measured
In a 2025 study, Microsoft researchers working with IBBIS and four DNA synthesis providers found that several production screening tools missed most AI-redesigned toxin sequences.
Founding team
Computational biologist with a PhD from the University of Arizona and experience conducting research for the U.S. Department of Energy. Leads Foldguard's structural screening and functional interpretability work.
Leads partnerships, fundraising, and strategy. BA from NYU, background in product development. Recently cycled 18,000 km from Portugal to Singapore.
Work with Foldguard
We'd like to hear from AI labs, DNA synthesis providers, and teams developing biological models. Get in touch to discuss screening or therapeutic design.
Get in touch