For years I told people I owned phagecocktails.com and had not built anything on it. That part is over. The site is live.
I want to be plain about what it is and why it exists, because the opportunity behind it is the largest and most neglected one I have ever called.
The one-sentence version
The most precise antibiotic ever discovered is a virus.
A bacteriophage is a virus that infects bacteria and nothing else. It finds one strain, replicates where that strain lives, and goes quiet when the strain is gone. A broad-spectrum antibiotic does the opposite. It kills across the whole community and takes the beneficial organisms with it. Antibiotics carpet-bomb. Phages snipe. phagecocktails.com is the case for building medicine around that difference.
Why this decade
Antimicrobial resistance was directly responsible for 1.27 million deaths in 2019, and associated with nearly 5 million, more than HIV and malaria combined. The O'Neill review put the trajectory at ten million deaths a year by 2050. Against that, the antibiotic pipeline has been close to empty for decades, because a drug you take for ten days and are told to use sparingly is a poor business next to a drug you take for life. We built a civilization on antibiotics and let the factory rust.
Phages are the thing in the cabinet with the right mechanism. They carry a century of clinical use: the Eliava Institute in Tbilisi has treated patients with phage preparations since 1923, and when Western medicine runs out of antibiotics, phages keep pulling people back one compassionate-use case at a time. Tom Patterson at UC San Diego in 2017. A teenager with a disseminated Mycobacterium abscessus infection in 2019. The therapy works. The system to deliver it was never built.
What is on the site
phagecocktails.com is a research and education platform, and it goes deep:
- Phage Therapy 101 and the 2050 crisis: the science and the stakes, from the ground up.
- Landmark cases: the rescues that demonstrate the mechanism, documented.
- A Build-a-Cocktail Lab and Find a Phage: interactive tools for matching a pathogen to the phages that kill it.
- Steal This Grant: fifty open-source, CC0-licensed, NIH-style grant proposals. Fork them. Run them. Put my name nowhere near them if you like.
- The flagship: a prophylactic cocktail to prevent necrotizing enterocolitis in premature infants by suppressing Klebsiella blooms.
Why I am giving the grants away
Fifty proposals, CC0, no strings. People ask why I would hand out ideas this valuable. The honest answer is that an idea sitting in my head saves no one, and I cannot run fifty trials myself. The bottleneck on this field is execution and capital, so the useful thing I can do is lower the activation energy for everyone who can run one. Take a proposal. Win the grant. Save the patients. I would rather the work happen than be credited for it.
The part almost no one is building
I have spent a decade formalizing microbiome signatures: the condition-specific patterns of which organisms are elevated and which are depleted in a disease. Connect that to phage therapy and the signature becomes a targeting system. If you know which taxa are driving a person's disease, you know which phages to load into the cocktail. You are no longer treating an infection in the abstract. You are removing, organism by organism, the configuration making this particular person sick, and leaving the rest of their ecosystem standing. That is precision medicine of a kind that is not yet on anyone's roadmap, and it is buildable now.
What this is, and what it is not
phagecocktails.com does not treat anyone. It is education and research, and it says so on every page that matters. The waitlist is interest registration, nothing more. If you are a clinician, a researcher, an investor, or a parent who has read the NEC literature and cannot unsee it, get on the list. If you need access to phage therapy now, the site points you to the compassionate-use resources.
There is a running summary of all of this at karenpendergrass.com/phage, and the long-form argument for why I bought the domain in the first place is here.
I am usually too early on the when. I have never been wrong on the what.