Research-stage · building the foundation model

The definitive map of cancer resistance.

Cancer care is still too often a high-stakes guess. Enso is building a causal AI platform that learns from histopathology, molecular profiles, and clinical context — and then tests the biology to turn insight into action.

“The secret of the care of the patient is in caring for the patient.”
Francis W. Peabody — a reminder that technology must serve humanity.

A platform built for actionability

Precision oncology needs more than prediction. It needs robust generalization, mechanistic grounding, and a workflow that respects both doctor and patient.

Multimodal foundation

We start with the richest signal: tissue. We then align histology with transcriptomics and proteomics, and later with clinical narrative and exposure history — so the representation learns what matters.

Causal validation loop

Our models propose resistance mechanisms and biomarkers; we validate key hypotheses in the wet lab (e.g., functional genomics screens), closing the loop from correlation to cause.

Clinician-centered

The goal is not to replace clinicians — it’s to give them time back, reduce uncertainty, and offer transparent evidence in a domain where decisions are rarely deterministic.

Roadmap

We build from robust foundations toward prospective validation. Each step is designed to produce measurable learning — and better care.

Now · Foundation

Histology foundation model

Train a long-context slide model on public cohorts (TCGA, GTEx, CPTAC), evaluate on external benchmarks, and build the infrastructure that makes scaling inevitable.

Next · Multimodal

Proteomics-aware learning

Add a proteomic regression objective on CPTAC to enrich the latent space with molecular signal. This is the first step from pattern recognition to actionable biology.

Then · Causality

Validation and pilots

Partner with leading cancer centers for pilot studies and prospective evaluation; validate key mechanisms with functional assays; translate insights into clinical and trial decision support.

North Star

A global map — for everyone

Resistance mechanisms vary across populations. We want to build with the world, not just for it — increasing representation, reducing bias, and returning value to partner institutions globally.

UMAP embedding space and spatial map (example visualization)
Example: learning structure in tissue. Left: embeddings form clusters; right: those patterns map back to the slide. (Illustrative visualization from our internal prototyping.)

Partnerships

We work with hospitals, cancer centers, researchers, and biotech to build models that are robust, validated, and useful.

Hospitals & cancer centers

Collaborate on data and validation. Partners can be recognized in publications, receive early access, and help shape a system designed for real workflows.

  • De-identified slide + outcome cohorts, with clear governance.
  • External validation and site-specific robustness analysis.
  • Priority access / partner pricing for pilots.

Pharma & biotech

Use the causal engine to de-risk trials, identify biomarkers, and uncover resistance mechanisms worth validating.

  • Patient stratification and biomarker discovery.
  • Mechanism hypotheses grounded in multimodal evidence.
  • Validation strategy aligned with wet-lab experimentation.

Team

We are an interdisciplinary founding group spanning medicine, machine learning, and experimental biology.

Luca Maria Menga

Founder · MD/BSc (Humanitas University & Politecnico di Milano, MEDTEC). Research experience across digital pathology, structural biology, and high-throughput wet lab (Harvard, Stanford/SLAC).

Matteo

Cofounder · Building scalable systems and product foundations for clinical-grade deployment.

Mattia

Cofounder · Physics & statistics background. Focused on rigor, evaluation, and modeling strategy.

Contact

If you are a cancer center, hospital, research group, or biotech team and want to collaborate, we would love to talk.

Important: Please do not submit patient identifiers or protected health information through this form. We can share appropriate secure channels during follow‑up.

By submitting, you agree to our Privacy notice.