ALP Bio / immune organoids + AI
Swiss startup ALP Bio raised €1.9 million for a platform where immune organoids and AI test in advance whether the body will start attacking a therapeutic antibody
On April 30, EU-Startups reported on an early round for ALP Bio from Schlieren. The company is building a system for antibody developers: it takes human immune organoids, measures the response to a molecule, and then uses AI to reduce immunogenicity risk. The round was led by 42CAP, joined by Venture Kick and private investors.
Antibodies have long been the workhorses of modern medicine: they are used to treat cancer, autoimmune diseases, rare disorders, and inflammation. The problem is that the body sometimes starts to see such a drug as a foreign intruder. In response, anti-drug antibodies, or ADA, appear: immune proteins that bind to the drug itself. They can bind the medicine, reduce its effect, alter how it moves through the bloodstream, or trigger adverse reactions.
That kind of failure often shows up late. The molecule has already gone through design, lab testing, manufacturing, animal models, and the first steps toward the clinic, and then the immune system starts interfering with the therapy in patients. A review of biologic drugs on PubMed states this plainly: almost all therapeutic proteins trigger an immune response, and high, persistent levels of antibodies against the drug can lead to a clinical loss of efficacy.
ALP Bio wants to move that check much closer to the start of development. On the company's website, the workflow is described in three steps. First, protein models assess antibody sequences and look for risky regions. Then the candidate is tested in tonsillar organoids — small models of immune tissue grown from human tonsils. These organoids are meant to capture B-cell and T-cell interactions that a standard sequence-based calculation can miss. After that, generative AI suggests how to rewrite the risky regions of the antibody while preserving its therapeutic function.
AI provides speed; the organoid provides biological feedback. Chief Scientific Officer Lucas Schaus describes ALP Bio's position this way: assessing immunogenicity requires measurements from living immune tissue alongside sequence-based computation. The company wants to train models on responses from human immune tissue and then feed that result back into molecule design. It is closer to an engineering loop: test, spot the weak point, rewrite, test again.
The round is small: €1.9 million. The company was founded in 2025, and the article relies on statements from the company and its investors; the platform's accuracy still needs to be validated with external data. According to Venture Kick, the money will go toward expanding the organoid platform, automation, higher throughput, and early partnership projects with antibody developers. This is an early claim for a tool: ALP Bio still has to show how accurately its system predicts immune responses in humans.
In AI biotech, the first step usually gets the spotlight: find a protein, design an antibody, propose a molecule. At Galux and GC Biopharma, that model looked like this: AI designs antibodies against autoimmune diseases, and a large biopharma company tests them in the lab. ALP Bio is targeting the next expensive bottleneck: why a good candidate breaks down on the way to the patient. A review of therapeutic antibody developability describes a similar shift: assessments of stability, solubility, non-specific binding, and other risks are increasingly being pushed into early candidate discovery because late-stage failure is simply too expensive.
Therapies for age-related diseases are already moving toward biologic medicines: antibodies, protein constructs, RNA, and cell and gene interventions. Each of these therapies has its own version of the same question: how do you deliver the effect into the body while preventing the defense system from attacking the treatment itself. The earlier a developer can see that risk, the less medicine looks like a series of expensive surprises in the clinic.