QAI Labs · NOVA-3 platform

Designed,
not screened.

Three programs. One quantum-enhanced platform. Eight to sixteen weeks from sequence to in-vitro PoC; in-vivo timeline partner-lab dependent.

Live in your browser — real ESMFold + Mol*, no login, no signup.

8–16 wk
Target · seq → in-vitro PoC
>60%
Target PoC success rate
1,200+
VH/VL pairs · 80+ targets
>3%
Modeled BBB brain exposure

Design targets and in-silico / modeled figures — not yet demonstrated in vivo. Library size is current. Measured results publish on Benchmarks as programs mature.

A five-layer stack for multi-specific antibody design.

Quantum-enhanced AI fused with a wet-lab data flywheel. Each layer feeds the next; each candidate carries a cryptographically anchored evidence trail.

01
Layer · Data

Module Library

1,200+ pre-validated VH/VL antibody pairs across 80+ therapeutic targets in oncology, neurology, and senolytics. Plus a curated library of linkers, hinges, and effectors. Each module is provenance-stamped, partner-IP-segregated, and schema-typed.
1,200+ VH/VL pairs 80+ targets Partner-IP carveouts
02
Layer · Compute

NOVA-Compute

AWS GPU for classical inference, IBM Heron r3 (ibm_boston · 156Q) and Quantinuum Helios (98Q, all-to-all) for CVaR-VQE / QAOA, Modal & Runpod for burst. Every job logged with backend, shots, calibration snapshot, and cost.
Quantum-hardware-explicit Simulator fallback Cost ledger
03
Layer · Model

Generators & Heads

Best-of-breed open foundations wrapped for multi-specific design: Boltz-2 (MIT license) and ESM-3 (Cambrian license; commercial use requires agreement with EvolutionaryScale), RFantibody (de-novo binders). AlphaFold 3 used for internal research benchmarking only under non-commercial terms. Three program-specific heads on top: ranking, structural confidence, in-silico Kd.
Three heads Mol* viewer Top-24 ranked
04
Layer · DBTL

Wet-lab Loop

Partner-lab adapters with IP-carveout enforcement. SPR/BLI, ADCC dose-response, in vivo efficacy. Every assay result flows back into the Module Library; flywheel learns +0.8% accuracy per program per cycle.
+0.8% / cycle SPR · BLI · ADCC In-vivo PoC
05
Layer · Evidence

NOVA-Evidence

Cryptographic provenance on every result. SHA-256 root committed to Bitcoin via OpenTimestamps. Inventor attestations filed at provisional time. Every claim — Kd, EC50, pTM, ADCC — reproducible from the bundle.
OpenTimestamps SHA-256 root Notary-grade

Three programs. Three provisionals.

Patent-defensible multi-specifics across oncology, neurology and senolytics — each on the same eight-to-sixteen-week path from sequence to in-vitro PoC; in-vivo schedule partner-lab dependent.

NOVA-3 · P1

CLDN18.2 × CD3 × 4-1BB
trispecific TCE

Gastric / GEJ adenocarcinoma. Tumor-restricted T-cell engagement with 4-1BB costimulation to address CD3-bispecific exhaustion. Provisional filed Q2 '26.

CLDN18.2 ECL1 — tumor restriction
CD3ε — TCR engagement
4-1BB — costimulation gate
In silico PoC PROV-2026-001
NOVA-3 · P2

TfR1 / IGF-1R BBB-VHH shuttle

Blood-brain-barrier penetration shuttle for CNS-restricted biologics. Bispecific VHH design optimized for transcytosis kinetics over receptor sequestration.

TfR1 — primary transferrin route
IGF-1R — capacity-additive
Cargo-agnostic format
In vitro PoC PROV-2026-002
NOVA-3 · P3

GD3 × IL-7R
senolytic bispecific

Carbohydrate-binding active space — the flagship moat. Senescence-cell targeting via GD3 ganglioside (Iltis et al., Nat Aging 2024). Second arm provides survival-pathway gating; specific co-target under wet-lab evaluation.

GD3 ganglioside — senescence marker
IL-7R — survival pathway
Quantum-corrected glycan binding
Lead candidate ranked PROV-2026-003

Evidence, not assertions.

On raw folding accuracy, AlphaFold 3 and Chai-2 set the bar; NOVA-3 wraps them. Our differentiation is upstream — multi-specific composability via the Module Library — and downstream — quantum-corrected refinement, cryptographic provenance, and a wet-lab flywheel. Every result is reproducible from an Evidence Bundle.

Positioning map — predict vs. compose, with provenance as the second axis QUALITATIVE PLACEMENT · NO METRICS · STRATEGIC VIEW WHERE WE PLAY PREDICT one structure COMPOSE multi-specifics from parts CRYPTOGRAPHIC PROVENANCE NO PROVENANCE General folding models AF3 · Chai · ESMFold Classical screening / CRO workflow NOVA-3 compose + signed provenance LEGEND Folding models Classical screening NOVA-3 our position Target space Position is qualitative — no axis is numeric.
Figure — Where we play. Two axes that matter: do you predict a single structure or compose a multi-specific from validated parts, and can you prove how the design was made. General folding tools and classical CRO screening cluster in the lower band — useful, but unsigned and structure-first. NOVA-3 sits alone in the upper-right: compositional design with cryptographic provenance. Placement is strategic and qualitative; the axes carry no numbers.
Bundle IDEVB-P3-2026-0517-3F2A
ProgramNOVA-3 · P3 (GD3 × IL-7R)
CandidateCAND-P3-LEAD-08
SHA-256 root0x9c4b8e21a3f7d24c81b6e5f9a2e4c7d5...
OTS anchorConfirmed · block 944,517
Anchored at2026-05-17 14:22:08 UTC
InventorWang · Gao · Feng
AttestationPROV-2026-003 §B-7
ReproducibleYes · cmd attached

Multi-specific composability — unique.

Curated, schema-typed Module Library across binders / linkers / hinges / effectors. None of AF3, Chai-2, Boltz-2 or ESM-3 ships an equivalent. We compose; they predict.

Quantum active-space corrections — unique.

CVaR-VQE on IBM Heron r3 for carbohydrate / glycan active spaces. No claim of demonstrated quantum advantage in ground-state chemistry — we engage with the open question.

Cryptographic provenance — unique.

No surveyed competitor in the AI/quantum drug-discovery set has shipped OpenTimestamps-anchored Evidence Bundles. Pharma diligence gets notary-grade invention dates.

Built in Texas. Global by design.

QAI Labs Ltd. is a Delaware corporation headquartered in Sugar Land, Texas, building at the intersection of pharma business development, quantum hardware partnerships, and multi-specific antibody engineering — with collaborators across North America, Europe, and Asia-Pacific.

F
Dr. Daniel Feng
Chief Executive Officer

Former Medarex / Bristol-Myers Squibb. Led antibody engineering on three programs to Phase I/II. At QAI Labs, oversees company operations, the NOVA-3 platform, and US-facing partnerships including the IBM Quantum and Quantinuum integrations.

G
Dr. Xin Gao
Co-founder

Serial entrepreneur and clinical physician with deep multi-specific antibody design experience. Co-founded QAI Labs and leads platform development across the NOVA-3 module library, the design pipeline, and the three internal P1 / P2 / P3 programs.

W
Dr. Min Wang
Chief Scientific Officer · Quantum AI Lead

Inventor of the proprietary XFab platform for molecular simulation. Owns the quantum-augmented design layer of NOVA-3 — CVaR-VQE active-space corrections, QAOA assignment optimization, and the hardware-software interface to IBM Quantum and Quantinuum.

Eight to sixteen weeks. Three programs. One platform.

Pharma BD, partner labs, and quantum hardware collaborations. The first conversation usually starts with a P3 walkthrough and an Evidence Bundle review. Or submit a target antigen below and we'll scope a design in 48 hours.

or email partner@qailabs.co

We respond within 2 business days.