Biolojic Design

The design is the product. Biolojic uses machine learning to shape antibodies into functional switches, agonists, antagonists, conditional binders and multibodies that bind more than one target on a single arm, rather than screening for whatever binds. Those designs are what partners license and what the pipeline is built on. There is no separate asset that survives if the model is removed, which is why the company describes itself as computationally designing antibodies rather than discovering them.

Biolojic's engine computationally designs an antibody and the pipeline then carries it forward, which makes clear that designs are tested before they advance. What that pipeline does not spell out is the control around the model: where a scientist reviews or overrides a design, what the company does when one fails in the lab, and which steps proceed with no person involved. The oversight can be inferred from the fact that molecules get tested, but it is not laid out as a mechanism.

The approach is described and the technology is not identified. Biolojic says it uses machine learning to design epitope specific antibodies and multibodies, which conveys what the system does without naming a model, giving an architecture or parameter count, or characterizing the training data. Posters and technical reports are published on individual antibodies, but no version or update practice ties the design system behind a given program to a date.

The engine is presented as the company's own and no outside model provider is named as sitting between a target and a design. Beyond that nothing is established: no compute or hosting provider is named, no third party components or licenses are listed, there is no subprocessor list, and nothing describes what governs a partner's target data once it enters the system.

This is the one antibody design company in this group whose work has reached patients. AU-007, an antibody designed on the platform and developed by Aulos Bioscience, was described as the earliest computationally designed antibody to enter a human trial when it began dosing cancer patients in 2022. The company's own lead, TEV-325, a multibody partnered with Teva, entered investigational new drug enabling studies for atopic dermatitis and asthma in 2025, and a multi target discovery collaboration ran with Merck KGaA.

What the record lacks is a published method. The public evidence is conference posters and technical reports rather than peer reviewed papers on how the platform designs, and the company's own programs are preclinical, with the clinical milestone carried by a partner's molecule rather than documented here with its trial readouts.

The platform works on antibody and target data, not patient records, so protected health information does not enter the picture. For a partner the live question is what becomes of its target data inside the system: whether it is retained, fed back into training, or isolated from other programs. Biolojic publishes no answer, so that handling is undescribed.

Biolojic handles antibody and target data rather than patient records, so United States health privacy rules do not reach its work. The site states no business associate position and publishes no privacy document that covers the design platform.

Partner discovery programs put another company's target data inside Biolojic's systems, and a partner's security team reviewing that arrangement would expect something independent to read: a SOC 2 report, an ISO 27001 or HITRUST certification, or a trust portal. Biolojic names none of these and runs no security page, so how that environment is protected is not attested anywhere public.

No device claim is made, and the company is scoped to match. Biolojic designs antibody drugs, so the regulatory position belongs to the molecules rather than to the design platform. AU-007 carries its own investigational new drug status through Aulos as its sponsor, and the company's own programs are preclinical or entering enabling studies, so there is no medical device here to clear.

What Biolojic shows about its model is a set of success stories: posters and a technical report walking through individual antibodies the platform designed. A buyer judging governance needs something different, an evaluation of how reliably the model succeeds or fails, a benchmark, an audit, or a responsible AI policy setting rules for its behavior. A worked example of a design that came out well is not that evaluation, and none of those governance artifacts is published.

A partner licensing a design wants to know what backs it when it does not work out. Biolojic publishes no hit rate or success rate carrying a method and a denominator, so there is no error figure to anchor expectations, and no guarantee, indemnity or correction route comes with a design. That allocation of risk is written into the private collaboration agreements rather than stated in public.

What the platform delivers is antibody designs, not software a buyer installs and connects, so there is no clinical record surface involved, as one would expect of a design company. The research side is equally self contained: no electronic lab notebook, laboratory information management system or research data platform is named as connected, and no connector is offered to a partner who would want designs to feed their own systems.

The platform runs for Biolojic rather than being deployed to anyone, so there is no tenancy or hosting model to evaluate. Location is implied by the company's base in Israel rather than committed, and nothing published names where partner target and design data rests or how one collaboration's data is kept apart from another's.

Biolojic earns through its own pipeline and its partnerships rather than selling access to the design engine, so no price appears. Its deals with Teva and Merck KGaA are named without public financial terms, which lets a reader see how the company works without being able to size the cost of anything.

Coverage is named with programs behind part of it. Biolojic works in autoimmune disease, where TEV-325 targets atopic dermatitis and asthma, and in cancer, where BD200 is a multibody drug conjugate and AU-007 reached a trial in cancer patients. The antibody types are named too: agonists, antagonists, conditional binders and multispecific multibodies. What is not established is broader clinical reach, since only the partner carried AU-007 into a trial and the company's own programs are earlier.