The MANIFOLD Platform

The frontier platform that learns the universal laws of molecular recognition to design modular protein therapeutics.

LLM Multi-Agent System: Toward Autonomous Biological Discovery

Interactome foundation model

A suite of models including foundation models, generative energy models and discriminative binding models that learn the topology of molecules. We unify sequence, structure and physics cognition to traverse biology’s high dimensional design space.

Novel Synthetic Biology Foundry

Innovative massively-parallel reporter assays that screen millions of AI-designed molecules per cycle.

Massive-Scale Proprietary Data Corpus

A corpus of hundreds of millions of binding interactions across disease-specific targets. Our rapidly growing dataset includes the "hard negatives" and cross-reactive interactions crucial to designing potent and safe molecules.

In a self-learning system, these capabilities converge to expand our system of biological cognition.

From Model to Molecule

Experimental Abundance

Novel synthetic biology techniques generate task-specific data for model training. All AI-designed molecules are experimentally validated in high-throughput assays. Autonomous AI is steered by high diversity lab data towards the next experiment.

MANIFOLD Model Suite

The MANIFOLD model suite decodes the rules that govern all molecular interactions.

A suite of models including:

  • LLM multi-agent system
  • Interactome foundation model
  • Generative energy models
  • Discriminative binding models
  • Structural and topological models
  • Affinity prediction and generation
  • Specificity and off-target prediction

Biomolecule Design

Our models design molecules that wouldn't be possible without generative biology

Using multi-conditional logic gates, we program molecular recognition:

  • AND gates: condition towards multi-specificity
  • OR gates: Relax restrictions to expand the design space
  • NOT gates: Avoid off-target toxicities by design

This is a fundamental shift from empirical screening to programmable biology.

From Recognition to Cognition

First Application: TCR T Cell Engagers

Targeting intracellular proteins is the next frontier in solid tumour oncology and cell-specific depletion in immunology.

We're developing a pipeline of T cell receptor T cell engager (“TCR TCE”) therapeutics against challenging targets.

By harnessing MANIFOLD to navigate the complete recognition landscape, we design molecules with properties that nature never evolved:

An agentic AI system to scale biological discovery

Synteny’s self-improving ML backbone learns from every experiment, decision, and human insight. It evolves autonomously to explore new directions and optimise computation across the discovery process. We call this our intelligent substrate, with models-in-the-fabric.