>FASTA_9
Heritable Agriculture’s Googlification of Agriculture, Olipop and David’s Biotech Lemonade Stand, Arc Institute’s Evo2 is the GPT of bio, Maverick Metals raises $19 million, A poet's way of growing
Heritable Agriculture’s Googlification of Agriculture
Born at Google X, Alphabet's Moonshot Factory, Heritable’s mission is to dramatically reduce the cost and time of plant breeding (not gene editing just yet), starting with under-optimized crops like berries, avocados, oats, barley, rye, bok choi, and chickpeas, as well as forest trees like loblolly pine. Their customers include Italian strawberry producer CIV and forestry supplier ArborGen.
Traditional breeding relies on plant geneticists’ choice of genes and plants to cross, field trials that can take decades, and uncontrollable environmental variables. Heritable uses AI to analyze and match googles of transcriptional, weather, and phenotypic data at an in-vitro or in seedling stage, to select the most impactful genes for a desired trait, under certain environmental conditions.
They’ve developed systems to collect and freeze hundreds of samples in liquid nitrogen and have processed data from 14,000 samples spanning seven different crops, which may include plant height, biomass, bud density, leaf shape, drought and insect resistance, number of kernels, bitterness, and photosynthetic capacity.
Olipop and David’s Biotech Lemonade Stand
A few years ago, I thought my life’s mission was to start a David kind of biotech startup, one that would defeat not one but one thousand Goliaths. Today, I think it wouldn’t be so bad if v1 David, one that was perhaps as simple as a lemonade stand, became friends with one of those Goliath guys — LifeSci VC and others have been sharing crispy pharma M&A charts for years but… what about non-pharma, science-driven, profitable companies?
OLIPOP caught my attention as a brand even before their most direct competitor poppi got famously acquired by PepsiCo for $1.9B a few months ago — Both OLI and poppi are “functional beverages” packaged in colorful cans that claim to take care of your microbiota, have low-sugar content. Maybe because of the myriad of pages explaining the benefits of each of their prebiotics, OLI gives more “the real thing” vibes — A guilt-free, pleasure-maximizing product for the conscious-maximalist.
Indeed, CEO of Coca-Cola nutrition Becca Kerr, says they’ve been paying close attention to the functional beverages space as it’s giving consumers more chances to try different things at different times in the day. Meanwhile, OLIPOP founders Ben Goodwin and David (🫣) Lester, who’ve already exiting their first probiotic soda Obi, have taken OLIPOP to sell over $400 million in 2024, and recently raise a $50M Series C too.
Whether or not OLI joins Coca-Coliath, they’ve proven a point: your product doesn’t need to be cutting-edge biotech to sell a biotech-driven value prop or insight — Educate consumers about a simple facts like eating well and create the simplest version of a product that makes that sexy within the current zeitgeist — That’s how you build a Biotech Lemonade Stand.
Arc Institute’s Evo2 is the GPT of biology
If there’s a Google for biology, our friends at Arc have built the GPT — Evo2 is a genomic foundation model that can design functional DNA, RNA, and proteins up to 1M kb, given a short DNA prompt. Unlike blackbox AIs, Evo understands genomic features like exon–intron boundaries, TF binding sites, and protein structural elements. It can then predict the functional impacts of coding and noncoding genetic variations, without alignment or fine-tuning.
Instead of treating all inputs equally, their StripedHyena 2 model architecture processes parts of the sequence in parallel using three different types of convolutions, which saves computation and helps it scale to million-base sequences. It’s also faster and uses less memory than transformers, as shown in the throughput scaling graph.
By using inference-time search, researchers can also design chromatin location and length into their sequences using Evo2. This technique involves keeping track of multiple possible output paths, and steering to a desired outcome. For the first time in bio foundational models, the authors demonstrated that increasing data during inference leads to improved performance for complex structures.
Arc cofounder Patrick Hsu has stated that Evo2 represents a key moment in the emerging field of generative biology, as the models have enabled machines to not only read and write, but also think in biology. Though the density of coding regions is still lower than in nature (which they’ll improve in future iterations), Patrick envisions Evo as infrastructure for a future Biological Appstore.
So do I.
The time for next-gen biotech creators to design and build biotech startups from their bedrooms is getting closer, if it’s not already here. IT’S TIME YOU UNLEASH YOUR CREATIVE BIOTECH POWERS — What will you grow?
Maverick Metals raises $19 million seed
Eric is one of those next-gen. I found his talk at the Texas Institute Leadership & Business Management particularly inspiring — The guy had been using enzymes to break down super toxic compounds like VX at the US Army Edgewood Chemical Biological Center when eureka struck him: what if we used enzymes to break down mining ores too?!
In two years, he rented a lab in the middle of the pandemic, used AI-guided evolution to accelerate the optimization of extremophile-derived enzymes, proved they could degrade ores more efficiently than traditional chemicals, quit his full-ride pass to med school, started and accelerated Maverick Metals through YC, and has now raised more a $19M seed round.
LithX, their extraction technology, boosts copper recovery from low-grade ore, tailings, and smelter slag without using harsh acids (which are costly to manage) or high heat (since their enzymes can operate at near-ambient conditions). This translates into higher recovery rates, lower operating costs, and lower environmental impact for mining companies like Euro lithium + Borates, Idaho strategic resources, and Puro Litio Brazil, whom they’ve now partnered with.
Unlike traditional biomining, LithX is cell-free (which makes it easier to handle), doesn’t produce sulfuric acid, and can be be applied in-situ. Soon they’ll be using it for the extraction of more critical and strategic metals like lithium, rare earth elements (REEs), molybdenum, and gold. Check out my extensive post to learn why I think this bioapp is a gold mine, in the business sense of the word ;)
A poet’s way of growing
Quantum mechanics and ecology have proven that our conception of the universe as a machine of separate and isolated parts doesn’t make up for the full puzzle of reality. If the proposal to compartmentalize and engineer biology is similarly limited by its masculine and minimalistic values, what might its female-maximalist counterpart be?
Just as color is not painting and notes are not music, molecules are not life. A poet of the living is different to a bioengineer in that she understands that she is one with the universe, not a master of it. She listens to her heart, who sings no differently to nature. She intuitively creates biology, because she is biology.
Living poetry is not reductionist, but generative. It’s not at odds with modern society; it adapts on its own terms, without surrendering to its core. Where prose expresses raw information, poetry shares deeper secrets through a veil, which leaves room for the imagination. Its beauty is not violent or intoxicating; we carry it along with us, silently, and meet with her again in dreams.
This is great! Happy to be a new reader here 🙌🏼