Abtique

The design is the product. A customer gives a target and the platform returns AI designed antibody sequences, and that is the entire service. The in silico screening heritage of EVQLV, picking strong candidates out of very large computational pools, is the same bet: the model does the work, and the lab validates the result. Nothing remains to sell if the model is removed.

Abtique returns designed sequences that have been independently validated in the lab, so a customer gets more than a raw model output. The structure around the model is not described: where a person reviews the designs, what happens when validation fails, and which steps run automatically. The validation reads as a backstop the service runs rather than a rule it spells out.

The technology is referred to by a name, HyperBind, and not explained. The site conveys that AI designs the sequences without naming a model class, giving an architecture or parameter count, or characterizing the training data, and no version or update practice is published. A reader learns that there is a model and little about it.

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

The service rests on a validation claim and some case studies rather than a checkable result. Abtique says the designed sequences are independently validated in the lab and returned in about two weeks, and shows case studies, but it names no current pharmaceutical partner, no clinical program and no outcome figures a reader can test, and publishes no peer reviewed work. As EVQLV the company ran an early antibody collaboration with ImmunoPrecise against SARS-CoV-2, which is history rather than a current result.

A customer submitting a target to Abtique is handing it to a hosted service, so the natural question is what happens to that target afterward, whether it is retained, used to train the model, or kept apart from other customers' work. The site does not say. No patient data is involved, so protected health information does not come into it, but the treatment of a submitted target is undescribed.

No business associate status is stated and no privacy document reaching the service is published. Abtique works with target and sequence data rather than patient records, so a HIPAA posture is not expected, and none is stated.

Customers submit their own targets to Abtique's hosted service, so an independent attestation is what a security reviewer would want: a SOC 2 report, an ISO 27001 or HITRUST certification, or a trust portal. The company has terms and a privacy policy but none of these, and no security page, so how the environment is secured is not attested in public.

No device claim is made, and the product is scoped to match. Abtique returns antibody sequences for research, so there is no clinical or diagnostic claim and no clearance to hold. Any regulatory position would belong to whatever a customer eventually develops from a sequence.

Abtique leans on validation, the point that each returned sequence is checked in the lab. That speaks to individual designs, not to how reliably the model performs across many attempts, which is the governance question. No benchmark, audit or responsible AI policy is published to answer it.

When a designed sequence does not work, the question is what the service does about it. Beyond the validation step it already runs, Abtique offers no stated recourse: no hit rate or success rate with a method and a denominator, and no guarantee, indemnity or correction route attached to a design.

A customer uses Abtique as a web product rather than integrating it with anything, so there is no clinical record surface, which a research design service would not need. No research system of record is named as connected and no connector or application programming interface is offered, so the designs come back through the product rather than into a customer's own systems.

Abtique is a single hosted web service, so there is a clear delivery model but no stated options or isolation to evaluate. Location is implied rather than committed: nothing published names where a customer's target and the designed sequences rest, or describes separation between customers on the shared service.

A buyer can size and start without a sales call, which is rare in this field. The site carries a pricing page, a how it works walkthrough and a way to request designs directly, and the unit is clear: antibody sequences for a named target, returned in about two weeks. The one gap is that the pricing page does not lay out full figured tiers, so a reader sees that the service is priced and roughly how it is charged, without a complete rate card.

Coverage is stated as a general capability rather than a named set. Abtique offers AI designed antibody sequences for research against a customer's target, which is broad, but no particular target class, disease area or validated program is named to show where the service has been demonstrated.