Axelyf.

Here is a thing about RNA medicine that sounds like a technicality but is actually the whole game: the molecule almost never fails. You can design a piece of messenger RNA to make a cell produce basically any protein you want, or a gene-editing payload to snip a specific piece of DNA, and the biochemistry will, in a test tube, do what you asked. The problem is the test tube. Bodies are not test tubes. Bodies are enormous, wet, defensive, and mostly uninterested in ferrying your delicate RNA to the exact cells that need it. So the RNA gets chewed up, or filtered out, or dumped in the liver - which is where nanoparticles tend to go whether you invited them or not.

This is the part everyone in the field calls "delivery," and it is the part that Axelyf, a small biotech headquartered in Brookline, Massachusetts, with its chemistry lab in Reykjavik, has decided to make its entire reason for existing. The framing matters, because most biotech companies are organized around a disease or a molecule. Axelyf is organized around a bottleneck. The founders came out of Moderna - the CEO, Örn Almarsson, ran drug delivery there during the period when Moderna was, you may recall, delivering an RNA payload to roughly the entire planet - and they left with a specific conviction: the payload is the easy part, and the vehicle is where the value and the difficulty both live.

The rate-limiting factor is delivery. The challenge is determining where to direct them and how to deliver them.- Örn Almarsson, Co-founder & CEO

The vehicle, in this case, is a lipid nanoparticle - a tiny fatty bubble that wraps around the RNA and smuggles it into cells. The most important ingredient is the "ionizable lipid," a molecule that is neutral in the bloodstream and turns positively charged once inside a cell, which is the trick that lets the particle release its cargo. Get the lipid chemistry right and you get potency, safety, and - the holy grail - the ability to steer the particle somewhere other than the liver. Get it wrong and you get toxicity, immune reactions, or a very expensive way to treat a hepatocyte that didn't need treating.

The AXL library

Axelyf's answer is a proprietary collection called the AXL lipid library: four distinct series of ionizable lipids, built through medicinal chemistry and consolidated in 2024 when the company acquired the assets of a firm called 76Bio. This is a slightly unusual origin story - a lot of platform biotechs grow their library organically over years, and Axelyf essentially bought a running start. The library supports both gene disruption (turning a gene off) and gene insertion (adding one), and it comes with surface chemistries meant to send particles to specific tissues rather than defaulting to the liver.

The headline data point, and the one Axelyf is happy to repeat, is a mouse study. In the liver, its AXL nanoparticles reportedly achieved four to five times higher gene-editing efficiency than LP-01 - a benchmark lipid used in gene-editing programs that are already in late-stage human trials at companies like Intellia and Verve - at equivalent doses, and did so without evidence of liver toxicity, and with reduced immune stimulation. Now, mice are not people, and every biotech's mouse data looks fantastic right up until the primates weigh in. But there is something bold about a seed-stage company benchmarking itself against drugs already in Phase 3. It is the corporate equivalent of a rookie asking to be measured against the starters.

We're not chasing science fiction, we're contributing real solutions.- Örn Almarsson

The AI named ANNA

The other half of the pitch is software. Designing lipids is a combinatorial nightmare - the number of possible molecules dwarfs the number you could ever synthesize and test. So Axelyf built ANNA, which stands, with admirable literalism, for Artificial Network for Nanoparticle Assessment. It is a machine-learning model that predicts how a given lipid structure will perform, letting the chemists prioritize which molecules are worth making. Axelyf says ANNA outperformed existing models when benchmarked against datasets published in Nature Biotechnology in 2024. This is the fashionable part of the business - "AI-driven drug design" is a phrase that raises money - but in fairness to Axelyf, lipid optimization is a genuinely good fit for machine learning, because the problem is exactly a search over a large chemical space with expensive, quantifiable experiments at the end.

What you can actually do with it

So what is the product? Two things, running in parallel. First, Axelyf is building its own drug candidates: AX-004, an in vivo CAR-T program aimed at solid tumors - the idea being to reprogram immune cells to fight cancer while they are still inside the body, skipping the expensive cell-manufacturing step - and AX-003, aimed at an autoimmune indication using liver-directed delivery. Second, and this is the part that makes Axelyf a "platform" rather than just a drug company, it wants to license the AXL library to other RNA-medicine developers who have a good payload and a bad vehicle. If you are a gene-therapy company whose editor works beautifully in a dish and disappointingly in an animal, Axelyf's proposition is: bring us your payload, we'll bring the delivery.

That dual model - own some drugs, rent the platform to everyone else - is a sensible hedge for a company with about fifteen employees and a $2.6 million seed round, which by biotech standards is a rounding error. The money, raised in August 2025 and extendable to $4.1 million, came entirely from Icelandic investors: Brunnur Ventures led, with Omega and Silfurberg alongside. The plan is not to do everything at once but to buy the next proof point - move the lipid library from mouse studies into non-human primates, and develop the surface chemistries that would let the particles finally, reliably, go somewhere other than the liver.

If that works, the interesting thing about Axelyf won't be any single drug. It'll be that they were right about the boring part. Delivery has been the quiet tax on RNA medicine for a decade. A company that lowers it doesn't get a molecule - it gets a toll booth on everyone else's.