Sovereign, autonomous research for molecular discovery

AI-accelerated research in a secure environment

DeltaWaveOS connects scientific reasoning, computational design, secure distributed compute, data, and laboratory execution in one environment.

How it accelerates

Scientists can delegate work to automation: literature reviews, producing data and analysing it. Scientists have limited hours, but agents can work relentlessly and take the weight off their shoulders.

Where it runs

Different scales, different deployments, different solutions. DeltaWaveOS can run on your own machines, in your cloud, or fully offline. Models, data and results stay inside your boundary.

How it learns

DeltaWave runs its own post-training infrastructure and closed benchmarks in a self-improving environment. Every company can get its own model, post-trained on its recorded runs and run securely on its own premises.

Harness

One question in front; behind it, agents that reason, run sealed code and GPU science, build on each other's evidence, and re-plan as results arrive.

Infrastructure

Start on a single compute instance; add any available hardware, at one site or many, and PCP places every job where it can run. Each lab keeps its own knowledge, and shares only what it chooses.

Training

Every run becomes a possible data point for post-training. The model learns its own moves, then practises in worlds we build with answers we know, and is tested on cases it has never seen. It replaces the version in use only if it holds up against it, so our customers have better models each day.

Selected research

Built from scientific practice.

Research across molecular design, biological representation learning, multi-omics, and auditable scientific agents informs the platform.

  1. 2026OmniPath: integrated knowledgebase for multi-omics analysisNucleic Acids Research
  2. 2025Protein language models for predicting drug–target interactionsCurrent Opinion in Structural Biology
  3. 2025Target-specific de novo design of drug candidate moleculesNature Machine Intelligence
  4. 2025An Auditable Agent Platform For Automated Molecular OptimisationarXiv
  5. 2025AI-driven discovery of antiretroviral inhibitors against poxvirusesCommunications Biology
  6. 2023SELFormer: molecular representation learning via SELFIES language modelsMachine Learning: Science and Technology