AI-Integrated Radiopharmaceutical Design

The drug-discovery lab,
under your fingertips.

Epitorix unifies the world's best AI design engines, an automated wet lab, and clinical translation on a single platform. Connect modules, press Run, and compress years of discovery into weeks — from de novo target to clinic-ready candidate.

Launch the platform See how it works

“Particles make life — when they are in the right place, at the right time.”

30+ AI design enginespeptides · binders · small molecules

~4-week wet-lab turnaroundsynthesis → binding → in vivo

2–3 years saveddiscovery to validated candidate

Our story

We didn't build another model.
We built the whole lab.

Drug discovery is fragmented. The best methods live scattered across hundreds of papers and repos; the wet lab is down the hall; the clinic is a continent away. Most platforms hand you a single model and wish you luck.

Epitorix is building the most comprehensive AI for radiopharmaceutical and pharmaceutical discovery. We unify 30+ high-impact AI engines from the literature — organized and ranked by our proprietary Epitorix AI Affinity module — to design de novo molecules against entirely new targets and established ones like PSMA and FAP, on a single canvas wired straight to an automated wet lab and a clinical network.

Design, score, validate in vitro and in vivo, and translate — without ever leaving the platform. Our internal programs span the full spectrum: from in silico designs to candidates confirmed by experimental affinity, and lead programs advanced through in vivo validation. The entire pipeline, and the entire body of published science, under your fingertips.

The problem

One approved drug hides behind 10,000 dead ends.

Traditional radiopharmaceutical discovery is a brutal funnel. Thousands of compounds are screened to land a single therapy — 10–15 years and billions of dollars — and most late-stage failures come from off-target toxicity and poor biodistribution found far too late.

Drug discovery

5,000–10,000compounds screened3–6 yrs

Preclinical

~250candidates advanced4–7 yrs

Epitorix replaces this first mile — thousands of designs generated & ranked in silico, in hours, for ~$0Below: the slow, expensive lab path — years & millions per molecule

Clinical trials

Ph 1 · 2 · 320 → 3,000 patients6–7 yrs

FDA review

1filing submitted0.5–2 yrs

Approved

1drug to market≈ 10–15 yrs total

1 in ~10,000compounds becomes an approved drug — after 10–15 years and $1B+. So you only ever pay the lab for candidates that already passed Epitorix's in-silico screen.

$50–100M

Discovery → IND

lead optimization + IND-enabling studies

$250–500M+

IND → approval

Phase I–III + manufacturing & CMC

5–8%

Oncology success rate

Phase 1 → approval, industry-wide

Where Epitorix accelerates

Same 15-year industry timeline. Epitorix collapses discovery and preclinical to under a year — reaching the clinic years sooner.

03691215years

Traditional

DiscoveryPreclinicalClinicalIND

~15 yrs

With Epitorix

Clinical≈ 8 yrs savedIND

~7 yrs

DiscoveryPreclinicalClinicalEpitorix AI

Preclinical compressed to ≤ 1 yearAI-guided design and de-risking before a single synthesis

Clinical success / failure, defined earlyscenarios mapped for FDA, cutting trial time and cost

Earlier clarity on where it workspredicted biodistribution shows where a tracer goes — or doesn't

How AI changes the math

Filter unviable ligands earlier

Predict protein–peptide binding better

De-risk off-target toxicity

Reduce synthesis & animal burden

Compress time to IND

How it works

Zero to clinic, on one continuous pipeline.

Every stage is a module on the same canvas. Snap them together, press Run once, and your candidates flow from idea to validated, clinic-ready therapeutics — no handoffs, no lost years.

01 · Design

Discover & design

Drop in a target structure. Chain 30+ high-impact AI engines from the literature — organized and ranked by the Epitorix AI Affinity module — to generate peptides, antibodies, nanobodies, enzymes, miniproteins, small molecules and even DNA/RNA binders de novo.

02 · Evaluate

Evaluate & rank

Score binding affinity (Kd), filter by developability and subcellular target, predict binding sites — or just chat with the AI to brainstorm and surface the best candidates.

03 · Validate

Wet-lab validation

Connect your data, drag in the Lab module and press Play. We synthesize and test your molecules in vitro and in vivo — full data and analysis back in ~4 weeks.

04 · Translate

Clinical & funding

Academic radiologists and nuclear-medicine specialists help push candidates toward trials — and we fund and co-develop the most promising programs with you.

The discovery engine

One target in. A ranked shortlist out.

Epitorix turns a single target into thousands of de novo binders — any modality, any target — then ranks them with the Epitorix AI Affinity module, so you walk into the lab with the few candidates most likely to work.

Any target

Drop in a structure or sequence — a validated clinical target, an entirely new one, even a small-molecule target.

30+ AI engines

30+ high-impact, peer-reviewed AI engines from the literature, plus our own frontier models — all in one canvas.

Any modality

Peptides, antibodies, nanobodies, enzymes, miniproteins, small molecules, DNA & RNA — designed de novo.

1,000s of binders

Generate thousands of candidate binders across the full design space in a single run.

Ranked shortlist

The Epitorix AI Affinity module ranks the full set, compressing thousands to a small, testable shortlist.

Epitorix AI Affinity

Boltz-2-level affinity. A million times faster.

Our Affinity module predicts binding at Boltz-2-level accuracy — but runs roughly a million times faster, screening billions of small molecules in hours, not years. Every candidate is ranked before a dollar is spent in the lab.

≈10⁶× faster than Boltz-2Billions of molecules / hours

Radiotheranostic isotope optimization

²²⁵Ac¹⁷⁷Lu⁹⁰Y¹⁶¹Tb

Small moleculesKd / Ki predictionPearson r0.68

R² 0.45vs. experimental Kd

PeptidesKd predictionPearson r0.83

R² 0.69vs. experimental Kd

Correlations on held-out experimental affinity. Same accuracy tier as all-atom methods — at a fraction of the compute.

30+ published, peer-reviewed methods — unified, organized & ranked in one platform

◆RFdiffusion◆ProteinMPNN◆AlphaFold3◆Boltz-2◆ESMFold◆RFpeptides◆BindCraft◆Chai-1◆ColabDesign◆PepMLM◆BoltzGen◆LigandMPNN◆RFantibody◆Genie 2◆CycleDesigner◆Protenix

All-atom molecular dynamics trajectory of a protein–ligand complex

Real molecular dynamics

The AI Design Hub

Every state-of-the-art model. One canvas.

Epitorix unifies 30+ high-impact AI engines from the literature — alongside our own frontier models — into a single workspace, ranked by the Epitorix AI Affinity module, then scored with real molecular dynamics. Connect modules and press Run, or just chat: the AI brainstorms, generates, and ranks your highest-affinity candidates.

Modular by design — snap engines together, branch, and compare runs side by side
Real MD, not guesses — all-atom dynamics and affinity (Kd) scoring on every candidate
Ask-AI copilot — describe your goal in plain language and let the agent design for you

30+ AI engines1,000s binders / runKd affinity ranking

Open the platform

EEPI Epitorix AI · online

Hey EPI — design a peptide that binds PSMA for a ¹⁷⁷Lu therapy.

On it. Running 6 design engines → 1,240 candidates, then Affinity ranking…

Top pick · EPX-PSMA-04ranked #1Predicted Kd3.2 nMOff-targetlowChelatorDOTA-ready+ 4 more ranked · dossier ready

Love it. Order wet-lab validation.

Queued synthesis + binding assays — full data in ~4 weeks. I'll text you. 🧬

Message EPI…

Conversational AI · built in

The whole lab, in a text message.

No code. No single line to write. Not even a button to find. Tell EPI what you need — “Hey EPI, I need a peptide that binds this target” — and the integrated AI runs the entire Epitorix pipeline for you: design across 30+ engines, Affinity ranking, de-risking, even placing the wet-lab order. Just talk.

Just type — or talk — plain language, no code, no menus, no parameters to tune
Runs the whole pipeline — design → affinity ranking → de-risking → wet-lab order
From your phone — the entire discovery lab, in your pocket

0 lines of codeVoice or text24/7 design copilot

Meet EPI

Path-traced render of a protein–peptide binder complex from the Epitorix Studio

Rendered in the PDB Studio

PDB Viewer Studio

Magazine-grade molecular renders. Free, in seconds.

Drop in any structure and get a Blender-quality, path-traced render — the kind labs pay artists weeks for — generated on our GPUs in seconds. Spin it, annotate interactions, and export publication-ready figures and animations. All under your fingertips.

Cycles-quality output — GPU path-tracing with real materials, depth of field and lighting
Interactions on tap — H-bonds, salt bridges and binding sites, auto-annotated
Figures & animations — export magazine-ready stills and MD trajectory videos

Seconds to render$0 per figure4K export

Open the Studio

Receptor–ligand complex · glass surface + cartoon

Binding interface · publication-ready lighting

Ligand close-up · ball-and-stick detail

Pocket detail · path-traced, depth of field

Rendered with the Epitorix PDB Studio — Blender Cycles on GPU, straight from structure files.

Render one live — free, in your browser

6VXXSARS‑CoV‑2 spike 1IGTIgG antibody 1GFLGFP β‑barrel 1BNAB‑DNA helix 1Z8LPSMA (prostate)4INSInsulin 6LU7Mpro + inhibitor 2RH1β2 GPCR 1HHOHemoglobin 1UBQUbiquitin

What you get back

Every candidate ships with a complete de-risking dossier.

Before a single wet-lab dollar is spent, each designed peptide arrives scored end-to-end — affinity, off-target load, stability, toxicity, charge, clearance route and chelator compatibility — plus predicted human biodistribution.

PeptideAffinity pKdOff-targetStabilityToxicityChargeClearanceChelatorCyclic

PEP-1179.2 ± 0.4low (8)stablelow+1.1renalLys-8yes

PEP-2048.7 ± 0.6med (24)stablemedium+2.2renalLys-3no

PEP-3187.9 ± 0.8high (47)mediummedium0.0mixednoneno

Plus generative human-predicted biodistribution & normal-organ dosepasswatchflag

30+

AI design engines, unified

~80%

Ranked-shortlist hit rate

~4 wk

Average wet-lab turnaround

0→100

Target to clinic, end-to-end

Automated 8-channel liquid handler dispensing into a 96-well assay plate

~4-week turnaround

Automated Wet Lab

Your molecule, in our lab. ~4 weeks.

Design is only half the story. Drag in the Lab module, connect your data and press Play — we place the order, synthesize your candidates, and run binding, in vitro and in vivo assays in our facility. You get full data and analysis back in about four weeks.

Drag, connect, Play — order wet-lab validation from the same canvas you designed on
Synthesis to in vivo — radioligand synthesis, binding, cell and animal studies
Full data + analysis — clean datasets and interpretation, not just raw readouts

~4 wk turnaroundin vitro + in vivoFull data + analysis

Send us a target

Predicted dynamic PSMA-PET biodistribution lighting a pelvic tumor lesion

Dynamic PET prediction

Clinical Translation & Funding

From bench to bedside, with people who've done it.

Our biodistribution models predict where a tracer goes — and our network of academic radiologists and nuclear-medicine specialists moves validated candidates toward first-in-human studies. For the most promising programs we don't just advise — we fund and co-develop, helping particles reach exactly the right place.

Predicted biodistribution — dynamic PET / dosimetry modeling before a patient is ever dosed
Path to trials — regulatory and translational guidance toward first-in-human
We invest — funding and co-development for standout science

Dynamic PET predictionFunded collaborations

Partner with us

Research & development

A pipeline of precision radiopharmaceuticals.

Programs advancing on the Epitorix platform — from AI design and experimental binding affinity through in vitro and in vivo validation. Each program sits at a different stage of preclinical maturity, with our lead programs now approaching clinical translation.

TargetIndicationModalityAI designExp. KdIn vitroIn vivoClinicalStage

FAPPan-solid tumor stromaFAPI radioligandIn vivo

LAMP1Solid & neuroendocrine tumorsTargeted radioligandIn vivo

GZMBImmunotherapy response imagingPeptide radioligandIn vivo

PSMAProstate cancer¹⁷⁷Lu / ²²⁵Ac peptideAI design

UndisclosedConfidentialTargeted radioligandIn vivo

UndisclosedConfidentialPeptide radioligandIn vitro

UndisclosedConfidentialTargeted radioligandExp. Kd

UndisclosedConfidentialMacrocyclic peptideIn vitro

UndisclosedConfidentialDe novo binderAI design

Programs span in-silico design through in vivo validation; the track shows each program's current preclinical maturity. PSMA is in silico (AI design) today. Undisclosed targets are under confidentiality.

Why Epitorix

Frontier models are brilliant. None are built for radiopharma.

The best open models predict structure or design generic binders. Radioligands have their own physics — affinity, stability, clearance and chelator chemistry — and that is exactly what Epitorix optimizes.

ModelWhat it does bestWhat's missing for radiopharma

Boltz-2MIT · Jul 2025Structural affinity for small-molecule predictionA structure predictor, not a design platform. Not peptide- or radiopharma-specialized.

BoltzGenStark et al. · Nov 2025Nanobody / miniprotein binder designNot specialized for peptides; wrong shelf-life and clearance kinetics for radiopharma.

Latent-XLatent Labs · Jul 2025All-atom binder designMacrocycle affinity is single-digit µM — too weak for radiopharma.

EpitorixEpitorix AI AffinitySingle-digit-pM peptides under 16 aa, optimized for chelator conjugation, protease stability and low off-targetPurpose-built for peptide radiopharma: high serum stability, nM–pM affinity, low off-target, biodistribution prediction.

Roadmap

A clear path from platform to first-in-human.

The platform is live and benchmarked against published baselines. The next milestone closes the loop in our own wet lab — then we scale to in-house radiochemistry, IND-enabling studies and the clinic.

Phase 1 · Months 1–6

Platform launch + first wet-lab validation

Live platform → first in-house validation loop

Public platform launch — hardened, multi-tenant web app
10–20 internal radioligand programs against validated targets
20 AI-designed peptides synthesized, confirmed by SPR + BLI
First peer-reviewed Epitorix AI Affinity calibration manuscript

Why Epitorix exists

The right molecule, in the right place, at the right time.

Mission

Build the most comprehensive AI for radiopharmaceutical and pharmaceutical discovery.

We pioneer the discovery of entirely new molecules — and better ones for known targets like PSMA and FAP — by unifying every state-of-the-art method with our own frontier models, then proving them in our wet lab and the clinic. So any scientist can take an idea to a validated, clinic-ready therapy years sooner.

Vision

A world where discovery is measured in weeks, not decades.

AI designs, robots build, and targeted particles reach exactly the right place at exactly the right time. Epitorix is the platform where that future runs end-to-end — from the first generated sequence, through in vitro and in vivo validation, to the patient.

Team & scientific board

Built by people who design, make, and treat.

A team spanning machine learning, radiochemistry and clinical nuclear medicine — advised by leaders across the field.

Founder & CEO

Founder

Platform, AI & systems

Radiochemistry Lead

Wet Lab

Synthesis & radiolabeling

Nuclear Medicine

Clinical Board

Dosimetry & translation

ML Research

Frontier Models

Binder & peptide design

Scientific advisory board

Academic RadiologyNuclear Medicine & TheranosticsMedicinal ChemistryTranslational Oncology

Join our board or team

Plans

Start free. Scale as you go.

Design on the platform from day one. Wet-lab validation, clinical translation and funded collaborations are scoped per program — talk to us.

Tier 1 · Community

Free for verified researchers

Light generative engines (PepMLM & more)
Studio renders
Community support
Research & academic adoption

Start free

Design it today. Have it validated in weeks.

Request access to the platform, or partner with us on a program — from de novo design through wet-lab validation and the clinic. We also fund and co-develop the most promising science.

Request access Partner with us

The drug-discovery lab,under your fingertips.

We didn't build another model.We built the whole lab.