NeuroPrecision-Engine

Precision through every barrier.

Most molecules bind. 94% never cross the blood-brain-barrier. We find the ones that do — in days, not decades.

BBB-first, not binding-first
Quantum-enhanced scoring
The 94% problem

The Blood-Brain Barrier is where CNS programs go to fail.

Molecules don't fail because they don't bind. They fail because they never arrive. We engineered around delivery — not around docking.

Read the full mission →
Mouse to human, solved

Cross-species prediction. The translation gap, closed.

A mouse hit isn't a human hit. We simulate across 10 species so the surprises happen on a server — not in a Phase I.

Human
Primary clinical target
Mouse
Standard preclinical
Rat
PK/Tox validation
Dog
Cardiac + Vet CNS
Pig
BBB device delivery
Macaque
Final translation gate
Marmoset
Small primate · CNS
Zebrafish
Larval BBB · transparent
Fly
Genetics · neurodegen
C. elegans
High-throughput aging
Explore the 10-species translation gap →
Veterinary CNS Division

Parallel discovery for canine cognitive dysfunction, epilepsy and anxiety — a faster regulatory path that generates dual-species validation data alongside the human pipeline.

We don't just find drugs. We predict success across species.

A closed-loop translational intelligence platform. 10,000+ CNS targets across 791 indications and 10 species models. From concept to clinic, every decision is computed, validated and ranked.

Sixty-five chances to fail fast

65 gates. Six engines. One decision.

Six engines, sixty-five gates, one ruthless principle: every candidate gets one chance to disqualify itself before it advances. Fail at gate twelve — not at clinical month thirty-six.

End-to-end pipeline flow

65 GATES · 6 PHASES
1–13 Target Biology Which target? 14–20 Binding Feasibility Can it bind at all? 21–28 BBB Delivery Can it cross? 29–38 ADMET & Safety Is it safe? 39–50 Physics Validation Does it really bind? 51–65 Translation & Dossier Mouse → human → top-10?
1
Which target?
Target Biology
G1–G13
2
Can it bind at all?
Binding Feasibility
G14–G20
3
Can it cross?
BBB Delivery
G21–G28
4
Is it safe?
ADMET & Safety
G29–G38
5
Does it really bind?
Physics Validation
G39–G50
6
Mouse → human → top-10?
Translation & Dossier
G51–G65

What survives each phase

indicative drop-off · CNS small molecules
From candidate pool From 10 000 candidates to 10 partner-ready leads.

Most pipelines spend their compute on docking — the gate that arrives in month thirty-six. NPE inverts the order: causality, BBB transport and ADMET come before any physics-heavy run, so the 90+% of molecules that were never going to make it are dropped on day one.

“Fail at gate twelve, not at clinical month thirty-six.” — every survivor has crossed the BBB and cleared off-target safety before a single docking pose is computed.
G1–G13
Target Biology initial candidate pool
10 000candidates
G14–G20
Binding Feasibility passes ligand feasibility
~600candidates
G21–G28
BBB Delivery passes blood-brain transport
~180candidates
G29–G38
ADMET & Safety clears off-target risk
~60candidates
G39–G50
Physics Validation binds with stable pose
~25candidates
G51–G65
Translation & Dossier mouse → human, partner-ready
10leads

Fail at gate twelve, not at clinical month thirty-six

economics of early failure

Classical CNS pipeline

FAILURE FOUND AT MONTH 36
  • In-vivo studies start before BBB transport is verified
  • Off-target liabilities surface in tox studies — after 18+ months of compound work
  • Mouse → human translation gap discovered in Phase I dose-finding
  • Each failed lead burns 3–4 years of program time
Cost per failed CNS leadUSD 50–200 M

NeuroPrecision-Engine

FAILURE FOUND AT GATE 12 · HOUR 3
  • BBB delivery is gate 21–28 — solved before any docking spend
  • ADMET + neurosafety at gate 29–38 — before physics-heavy compute
  • PBPK digital twin at gate 51–65 closes the mouse → human gap
  • Dead leads are stopped in hours, not in months
Median run, one disease~ 8 hours · GPU
Our mission

Make CNS drug discovery decidable on a server.

Built in Switzerland. Engineered around one conviction: the Blood-Brain Barrier is a solvable computational problem — and CNS doesn't have to mean a 94% failure rate.

94%
CNS drug failure rate, Phase I → approval
6% reach approval 94% never arrive
The problem

94% of CNS drugs never make it.

Most CNS molecules don't fail because they don't bind. They fail because they never reach the brain. The Blood-Brain Barrier is where billion-dollar programs go to die — usually in Phase II, after the chemistry already worked. The industry treats delivery as a late-stage surprise; we treat it as gate one.

Our solution

A closed-loop AI framework, CNS-first.

NeuroPrecision-Engine is the discovery stack engineered around the blood brain barrier from the first line of code. Real physics, real biology, CNS-trained machine learning, quantum-enhanced scoring where classical methods lose resolution. No LLM wrappers. Real numbers, every gate.

65
validation gates
6
engines
791
CNS indications
10
species
1Target BiologyG1–G13
2Binding FeasibilityG14–G20
3BBB DeliveryG21–G28
4ADMET & SafetyG29–G38
5Physics ValidationG39–G50
6Translation & DossierG51–G65
How we think

Three rules. One Engine.

Numbers, not narrative.
Real physics, real biology, real numbers. If we can't measure it, it doesn't ship.
Swiss-sovereign by design.
Built in Switzerland. Your IP, your data, your models stay under Swiss jurisdiction. Neutrality, applied to neuroscience.
Deep, not wide.
One problem: the blood brain barrier. The hardest one in drug discovery — and the only one we work on.
Two founders. One mission.

And then... us.

A science-and-technology founding pair. One Chief Scientist, one Chief Technologist — sharing one conviction: the blood brain barrier is solvable, and CNS deserves a discovery stack engineered for it from the first line of code.

Founded 2026 · Sursee, Switzerland
🧬
The Science

Neuroscience · Biology · BBB

Designs the 65-gate pipeline from first principles. Owns target selection, BBB biology, validation strategy and pharma relationships.

by Suvija
×
One
Engine
The Technology

GPU · Code · Infrastructure

Translates the pipeline into a live, GPU-accelerated system. Owns infrastructure, frontend, deployment, integrations and 24/7 operations.

by Tahnee
Suvija Münow-Suthakaran
Science
Co-Founder · CSO

Suvija Münow-Suthakaran

Chief Scientific Officer

Co-founder and platform architect. Designed the 65-gate pipeline from first principles and leads the science: target selection, BBB biology, validation strategy and pharma relationships.

Education
  • Specialization MSc in Experimental Biomedical Research, neuroscience focus — University of Fribourg (ongoing)
  • BSc in Biotechnology — ZHAW Zurich University of Applied Sciences
Tahnee Jessica Münow
Technology
Co-Founder · CTO

Tahnee Jessica Münow

Chief Technology Officer

Co-founder and technology lead. Translated the pipeline into a live, GPU-accelerated system and runs the engineering: infrastructure, frontend, deployment, integrations and 24/7 operations.

Education
  • BSc in Business Information Technology — Kalaidos University of Applied Sciences (ongoing)
  • Business Data Processing Specialist, Advanced Federal Diploma of Higher Education — Machine Learning specialization — ipso Bildung (ongoing)

The Blood-Brain Barrier isn't a mystery. It's an engineering problem.

Most CNS failures are failures of delivery, not of chemistry. We moved that question to the beginning of the pipeline — modelled through physics, biology, machine learning and quantum-enhanced scoring, gate by gate. Built in Sursee, Switzerland.

Let's build

Where the next CNS breakthrough begins.

Pharma. Biotech. Academia. CROs. Bring us a target, a disease, or a question — we'll show you the answer in days, not quarters.

📍
Location
Sursee, Switzerland
Response time
Within 48 hours

Get in touch

Tell us the indication or target class — we'll get back to you within 48 hours.

Thanks — we received your message and will get back to you within 48 hours.
The catalog

791 CNS indications. 10 species per pipeline.

All 791 indications carry human ICD-10 codes — that's the regulatory anchor. Each indication is also tagged with the 10 species commonly used to model it preclinically (mouse, rat, dog, pig, macaque, marmoset, zebrafish, fly, C. elegans, human). Filter by category, species or just type to search.

Loading 791 diseases…
Pipeline mode A

Start with the disease.

Pick one of 791 indications. The Engine harvests the targets, builds the compound library, and runs all 65 gates against that disease's biology — automatically.

1

Pick the disease

Type a name or ICD-10 — Glioblastoma, Alzheimer's, Rett syndrome, anything in the catalog.

2

Auto-harvest targets

Open Targets ≥0.4 association + STRING-PPI hop-2 + UniProt cross-refs build the target list — no manual seed needed.

3

Auto-build the library

ChEMBL bioactivity (pchembl ≥4.0) for the top targets, plus PubChem CNS focus set, plus DrugBank ATC N0x — deduplicated by InChIKey.

4

Run all 65 gates

BBB filter first, then ADMET, then docking, then quantum scoring — fail-fast at every step. Final dossier in days, not quarters.

Pipeline mode B

Start with the target.

Already know the protein? Type its symbol or UniProt ID. The Engine harvests every known ligand, generates new ones, and runs all 65 gates around that single target.

1

Pick the target

Symbol (EGFR, SNCA, GBA1, LRRK2, MAPT, BACE1…) or UniProt accession — anything druggable.

2

Resolve structure

UniProt → AlphaFold v6 + every PDB drug-bound template the target has. Class-aware: kinase, GPCR, ion-channel each handled differently.

3

Harvest + generate ligands

Pull every known active from PubChem and ChEMBL, expand chemical space with structure-aware generation, and dedup.

4

Run all 65 gates around the target

Same gate sequence as a disease run — but every result is anchored to your one target. Top-10 lead dossier ranked by ΔG, BBB, safety, and novelty.

The translation gap

10 species. One human prediction.

A drug that works in a mouse fails in humans 89% of the time. We close that gap by simulating across the entire evolutionary tree — and combining the signals into a single human-translation score.

Human
Homo sapiens
The clinical target. Every other species is judged against this one.
Mouse
Mus musculus
Default preclinical model — but P-gp, CYP3A and BBB-transporter expression diverge.
Rat
Rattus norvegicus
PK/Tox gold standard. Best-validated brain-MS data after human.
Dog
Canis familiaris
Cardiac (hERG-relevant) and veterinary CNS — canine cognitive dysfunction parallels Alzheimer's.
Pig
Sus scrofa
Closest large-animal BBB anatomy to humans. Used for device-mediated delivery.
Macaque
Macaca mulatta
Final translation gate before IND. Brain volume, transporter biology and cognition are nearest to human.
Marmoset
Callithrix jacchus
Small primate. CNS phenotyping at scale — faster reproductive cycle than macaque.
Zebrafish
Danio rerio
Larval BBB is transparent. High-throughput in-vivo CNS exposure imaging.
Fly
Drosophila melanogaster
Genetics powerhouse. Tau, α-synuclein and amyloid models at fly speed.
C. elegans
Caenorhabditis elegans
Mapped neuron-by-neuron. Ageing and neurodegeneration screening at extreme throughput.

How we combine them

1

Per-species PK simulation

PBPK with species-specific brain compartments + transporter expression (P-gp, BCRP, MRP1) gives a brain-Cmax for each animal.

2

Ortholog conservation

Sequence + structural conservation of the target between species; weight is downgraded where active-site residues diverge.

3

Pathway divergence

Reactome + KEGG cross-species pathway maps flag where compensatory biology breaks the model.

4

Bayesian fusion → TRS

Each species' signal becomes a posterior. We fuse them into a single Translation Reliability Score with a calibrated confidence interval — that's the number that goes in the dossier.

Privacy

Privacy Policy

Impressum

Impressum

Terms / AGB

General Terms and Conditions (AGB)

These General Terms and Conditions ("AGB") govern the access to and use of the website neuroprecision-engine.com (the "Website") and any non-binding pre-incorporation communication with the operators. They are based on Swiss law, in particular the Swiss Code of Obligations (CO / OR, SR 220) and the Swiss Civil Code (ZGB, SR 210).