FREQUENTLY ASKED QUESTIONS
Feromics’ Intelligent Design™ (ID) Platform is a proprietary, function-first single-cell system that integrates advanced immunology with AI-driven bioinformatics. It begins by isolating immune cells specifically selected for their functional relevance to a disease. The platform generates high-quality functional immunomics datasets, which power AI models for target discovery, biomarker identification, and translational applications in drug development and diagnostics.
Feromics improves the data layer, not the modeling layer. By supplying functional, disease-specific datasets, AI models start with clean, high-signal information. This leads to faster training, fewer required samples, and more reliable, actionable outputs for downstream applications.
Far fewer than traditional bulk approaches. Starting with cells already selected for functional relevance reduces noise, meaning high-confidence datasets can be generated with fewer samples.
Datasets include:
-
Immune cell activity in disease
-
Functional immune cell subpopulations
-
Signaling pathways driving immune response
-
Interactions between immune cells and disease biology
These datasets can be used for AI model training, target discovery, and biomarker identification.
-
Feromics’ patented microwell array technology allows high-throughput functional single-cell analysis, generating datasets at the scale required for AI applications, drug discovery programs, and large research initiatives.
Most single-cell platforms measure gene expression or protein content. Feromics measures cell function — what cells are actively doing in disease — at single-cell resolution and scale. This approach produces actionable, disease-specific insights that other platforms cannot deliver.
Until recently, technical limitations prevented isolating and analyzing immune cell function at single-cell resolution at scale. Advances in microfluidics and functional assays now make it possible, and Feromics’ platform was built specifically to leverage these innovations.
The platform is operational today, with published science, a patented technology, a $4.1M ARPA-H grant, and an active preclinical pipeline producing actionable discoveries.
Feromics partners across multiple phases:
• AI model development
• Drug discovery and preclinical research
• Translational programs for therapeutic or diagnostic development
Partnerships may include collaborative discovery, functional dataset access, or end-to-end translation of functional biology into pipeline assets.
Primarily a functional immunomics platform company. The Intelligent Design™ platform generates functional, disease-specific datasets that enable drug discovery, diagnostic development, and AI model training. When the platform identifies a target, Feromics can advance it in-house into a therapeutic candidate or diagnostic application.
• Published preprints demonstrating functional target discovery (e.g., KLRG1)
• Patent on proprietary microwell array technology
• $4.1M ARPA-H grant from the U.S. government
Functional immunomics at scale was technically impossible until recent advances in microfluidics and single-cell functional assays. Feromics built the first platform to generate this type of dataset, producing insights that simply did not exist before.
Feromics operates at the intersection of three growing markets: AI in drug discovery, immunotherapy development, and precision immunology. The platform can be applied to target discovery, biomarker identification, and therapeutic development.
Most AI models in drug discovery are trained on standard genomic data — bulk RNA sequencing, gene expression profiles, molecular markers. These datasets reveal patterns: genes that tend to appear together, markers that correlate with certain outcomes. But correlation is not causation. A gene that correlates with immune activity is not necessarily driving it.
Feromics AI is trained differently. Every data point in a Feromics dataset carries a verified functional label — a directly observed outcome attached to the molecular profile of the cell that produced it. A cell labeled "killer" was physically observed destroying a tumor cell. A cell labeled "exhausted" was observed losing function over time. The AI doesn't learn that certain genes tend to appear in certain contexts. It learns that certain gene programs caused certain functional outcomes — directly observed, physically verified, at single-cell resolution.
The difference matters enormously in practice. A correlational model finds patterns that may or may not be meaningful. A causal model learns what actually drives biological outcomes. Feromics AI produces predictions that are grounded in verified biology — not statistical association.
Not all donor cells are equal. For cell therapies — including CAR-T, CAR-NK, and Donor Lymphocyte Infusion — the quality of the starting material determines the quality of the therapy. A donor whose cells have high intrinsic killing ability, strong persistence, and resistance to exhaustion will produce a more effective therapy than a donor whose cells lack these functional qualities. Standard donor selection relies on compatibility matching — HLA type, health screening, basic blood counts. It does not measure what the cells can actually do.
Donor Intelligence is Feromics' AI-driven capability for identifying high-performance donors based on verified functional data. Using the Intelligent Design™ platform, we measure donor cells directly — observing killing behavior, synapse formation, serial killing capacity, and exhaustion trajectory at single-cell resolution. These functional measurements are fed into AI models that predict in vivo performance — how the cells will behave inside a patient, not just in a lab assay.
The result is donor selection based on what cells do, not just what they are. Partners receive a ranked functional profile of donor candidates — identifying the cells most likely to produce durable, high-efficacy therapeutic outcomes before manufacturing begins.
Feromics' competitive advantage is structural and compounding — built from four interlocking pillars that strengthen with every experiment run and every partnership engaged.
Function-labeled data that cannot be replicated. Feromics datasets pair verified functional outcomes with full molecular profiles at single-cell resolution. This data does not exist in any public repository. It cannot be downloaded, approximated, or inferred from existing atlases. A competitor wanting equivalent training data would need to build the experimental infrastructure from scratch and generate years of labeled data before approaching Feromics' current position.
Models that get smarter with every experiment. Each new dataset improves model accuracy and expands generalization to new cell types, diseases, and patient populations. This is a data network effect — the platform becomes more defensible the more it is used. Early partners benefit not just from today's models but from every improvement driven by every subsequent experiment Feromics runs.
Patented assays covering serial killing and persistence at scale. The microfluidic technology that enables Feromics' functional measurements is protected by patent. The patent specifically covers high-throughput measurement of serial killing and cell persistence at scale — the two most clinically valuable functional readouts in cell therapy development. A competitor cannot legally replicate these measurement capabilities without licensing from Feromics.
The Wet-Lab + AI Loop. The deepest and most durable moat is the integration itself — the seamless combination of biological data generation and AI model training in a closed, self-reinforcing loop. The AI informs experimental design. The experiments generate data. The data trains the AI. The trained AI suggests better experiments. Neither a pure biology company nor a pure AI company can replicate this loop. Feromics was built with both from the start.
Together these four pillars create a flywheel that widens Feromics' competitive advantage automatically — with every experiment, every dataset, and every partner.
Immunomics is the broad study of the immune system at a molecular scale — mapping the genes, proteins, and cells involved in immune function. It is comprehensive and valuable. But it describes what immune cells contain. It does not tell you what they are actively doing in the context of a specific disease.
Conventional immunomics approaches — including bulk RNA sequencing and standard single-cell genomics — measure molecular state. They capture a static snapshot of what genes are expressed or what proteins are present at a single moment in time. That snapshot cannot tell you whether a cell is killing a tumor, exhausting, persisting, or failing. It cannot tell you which cells are driving disease and which are responding to therapy. It cannot tell you which patients will respond to treatment and which will not.
Feromics measures function directly. Using our proprietary Intelligent Design™ platform, we identify and isolate immune cells based on what they actually do in the context of a specific disease — before any molecular data is collected. A cell labeled "killer" in a Feromics dataset was physically observed destroying a tumor cell. A cell labeled "exhausted" was observed losing function over time. That verified functional label is attached to the cell's full molecular profile at single-cell resolution.
The result is a fundamentally different kind of data — function-labeled, disease-specific, and verified at the source. Conventional immunomics tells you what immune cells contain. Feromics tells you what they do. That distinction determines the quality of every discovery, every AI model, and every clinical prediction that follows.