Skip to content

Products

Three assets. One system.

Axiom™, the design agent, proposes; Ark™ wet lab and cryo-EM measure; the data infrastructure turns every round of evidence into the starting point of the next. You state the design brief, the whole system designs or optimises, and you receive a traceable report.

One brief, and how it moves through the system.

From brief to report is a line; between design and optimisation is a loop that can repeat; every piece of evidence for an optimisation round is measured by Ark™ and settles into the data infrastructure.

Target, framework and goals — agreed, then frozen as the input for the whole campaign.
Target, framework and goals — agreed, then frozen as the input for the whole campaign.
Axiom™Design agent

Axiom™ proposes. Our scientists decide.

Axiom™ is an antibody-design agent: on a given antibody framework it generates CDRs and binding poses, scores them with multiple models and coded criteria, and optimises within a bounded number of rounds. It is not software you log into; it is the R&D engine abYcloud runs internally. Budget, production authorisation and final delivery are signed off by people.

  • A fixed procedure, provably without skipped stepsEntry to and completion of every step is decided by code, not by the model’s own account.
  • Criteria judge independentlyThe model proposes; the criteria layer scores and releases. The model cannot approve its own work.
  • Every candidate carries full provenanceFrom frozen input to final ranking, everything can be recomputed.
Trace of one runRunning
InputTarget and framework frozenBrief agreedDone
AuthoriseBudget and production lineSigned by a scientist; the agent cannot self-approveHuman sign-off
Round 1Generate → gate → scoreCandidates enter the optimisation loopPassed
Round 2Redesign → gate → scoreAgainst Ark™ measurementsPassed
Round 3Pose-consistency reviewCriteria hold; referred to a scientistReview
DeliverFull recomputation and rankingReport issuedHuman sign-off
Ark™Wet lab and cryo-EM

Every round converges on a measured structure, not a prediction.

Ark™ automated wet lab turns Axiom™ candidates into molecules and measures their binding; cryo-EM resolves the complex to near-atomic resolution. That measured evidence returns to the optimisation loop and settles into the data infrastructure. Ark™ is where the system’s evidence comes from, and where the agent’s improvement comes from.

  • Automated wet labExpression, purification and binding assays, scheduled by round.
  • Cryo-EM inside the loopAI-driven structure determination, > 70% lower cost.
  • Structure-level feedbackWhat returns is a measured structure, not a predicted score.
Density → structureResolving
Near-atomicresolution
>70%cost reduction
DATAData infrastructure and knowledge base

Gold-standard data, AI-ready.

Antibody–antigen complexes, cryo-EM structures, precomputed structural properties and developability metadata, unified as training-grade datasets. It is the ground Axiom™ stands on, and the layer that is thicker after every campaign.

  • Curated complexesAntibody–antigen complexes with precomputed properties.
  • Cryo-EM gold standardExperimental ground-truth structures from Ark™.
  • Developability metadataA first-class input to every scoring pass.
Knowledge-base layersAccumulating
Curated antibody–antigen complexesStructure + affinity + binding free energyGrowing
Cryo-EM structuresReturned by Ark™ every roundGrowing
Developability metadataManufacturability, humanness, liability sitesGrowing
Campaign criteria recordsVerdict and basis at every gateGrowing

The system gets stronger with use.

We understand this as recursive self-improvement at the system level: not an agent rewriting itself, but every campaign leaving the knowledge base thicker, the criteria sharper and the toolchain newer. The next Axiom™ starts higher.

  • 01The knowledge base grows with accumulated data

    Each round’s measured structures and binding data enter the data infrastructure and become the next design’s starting point.

  • 02Criteria keep calibrating on measurements

    Ark™ results keep correcting the scoring and release thresholds, bringing filtering closer to experimental truth.

  • 03The toolchain iterates with each campaign

    The weaknesses a campaign exposes become the next version of the tools and the procedure.

A little strongerevery roundCampaignArk™ evidenceKnowledge base growsCriteria and tools update

Working with us, in three steps.

  1. 01Brief

    Tell us the target, the existing framework and the goal; we confirm feasibility and freeze the input.

  2. 02Design and optimise

    The whole system runs: Axiom™ designs, criteria filter, Ark™ measures — repeated until it converges.

  3. 03Report

    Candidate ranking, criteria verdicts and full provenance — a report you can recompute.

Let’s talk about your target and design goals.

One conversation is enough to tell whether this system can take on your project.