Phylox uses AI to discover novel modes of action for the resistant weeds breaking modern agriculture.
Superweeds now shrug off entire classes of herbicides. Resistance is spreading across the world's biggest cash crops — and the incumbent discovery pipeline has run dry.
Protein language models now model plant targets directly — biology that used to be a black box is suddenly legible.
Molecular screening that once cost a fortune is now cheap enough to search billions of candidates in silico.
The incumbent pipeline has stalled. The chance to define the next mode of action is open right now.
We point the AI at the GLR3.4 binding pocket — a new, unexploited herbicide target class.
Protein-language models predict which novel molecules bind the pocket, across billions of candidates in silico.
Candidates are scored and shortlisted by predicted binding, selectivity, and synthesizability.
Top hits are confirmed in the wet lab — turning predictions into real, working chemistry.
Validated modes of action are licensed to the major agriculture companies that need them.
Weed control is a $40B global market. Roughly $10B is failing as weeds outrun existing chemistry, and no new mode of action has reached the market in 30+ years. Phylox is built to discover the next one.
Independent science researcher (TRP pain-receptor channels). 1st, IFT Nutmeg & PepsiCo Engineering Awards. 1st at HackPrinceton. Researcher at Brown.
3× Regeneron ISEF Grand Award winner. 1st at JSHS '25, National STEM Champion '26, 1st in Applied Technology. Researcher at Yale.
IEEE-published AI/ML researcher (youngest author). 3M Young Scientist & GENIUS Olympiad gold, Science. 3rd at JSHS '25.
A pre-seed round to take Phylox from validated targets to confirmed, licensable hits — capital goes straight into compound synthesis, wet-lab assays, and field validation.