01Lifabrica · October 2026

Ask a questionabout a livingsystem. Get backa simulation.

Omphalos builds executable models of biology — mechanistic, kinetic, inspectable down to the reaction. Lifabrica is how you reach them.

Product
Lifabrica
Launch
October 2026
Engine
L++ · fully kinetic, multi-scale
Commons
Life Code Foundation
Working with
KRIBBYonsei University College of PharmacyInSiliCoxFuriosaAIInstitut Pasteur KoreaMcMaster UniversityKRIBBYonsei University College of PharmacyInSiliCoxFuriosaAIInstitut Pasteur KoreaMcMaster University
02The product

Four things you can do the day you get in

01

Ask in plain language

Pose a biological question the way you'd ask a colleague, and get a simulation-backed answer — not an LLM guess.
02

Run the experiment yourself

Apply a mutation, a dose, or a strain swap and see the outcome before you touch a bench.
03

Inspect the mechanism

Open any answer and trace it down to the individual reaction, whenever you want to look.
04

A library you can build on

E. coli, A. baumannii, HIV-1, statin PBPK-PD, and more — validated models, so you never start from zero.
03Why it's different

A prediction you can open up

The difference between a model that sounds right and one you can audit is where the answer came from.

DimensionAn LLM answerOmphalos
Where the answer comes fromA statistical pattern in a training setA simulation built first-principles from biochemistry
When something's wrongIt can sound right — with no way to tellTrace it to the exact reaction responsible
Your dataThe model is fitted to itUsed to validate a model, never to fit it
CoverageA new tool per use caseMolecule to organism, one platform
04Scale

What one model actually holds

~0

reactions and processes in Virtual E. coli's genome-wide simulation

0

genes expressed genome-wide (K-12 MG1655)

7 & 28

sigma factors and two-component sensing systems modeled

05Method

How a model gets built

AI provides the hands and tools; the human expert is the control tower responsible for scientific judgment.

  1. 01

    Draft

    Generative AI assembles model components — reaction networks, parameters, compartments — from literature and open-source models.

  2. 02

    Inspect

    A domain expert reviews the logic, assumptions, and parameters directly in L++ before anything is accepted. Genuine authorship.

  3. 03

    Simulate

    The model runs. Plausible-sounding but structurally wrong reactions tend to surface here, visible in the output.

  4. 04

    Calibrate

    The model is tested against experimental or omics data it was never fitted to, and revised.

  5. 05

    Deploy

    The validated model becomes reusable infrastructure — ported, extended, or deployed into a new application.

06News

Recent

  • Mar 2026Omphalos signs MOU with FuriosaAI, KRIBB (KPEC), Yonsei University, and InSiliCox
  • Dec 2025Joint research agreement with Institut Pasteur Korea
  • Dec 2025Omphalos secures BARDA support for pathogen-agnostic antimicrobial strategy simulation
  • Oct 2025Omphalos wins award at the Paratus Digital Health Accelerator showcase
  • Jul 2025Omphalos selected for the BARDA-backed Paratus Digital Health Accelerator
  • Jun 2025Omphalos joins the NVIDIA Inception Program
07Questions

Asked and answered

No. There's no training corpus behind a model — the inputs are knowledge: biochemistry, enzyme kinetics, published regulatory-network structure. Multi-omics data is used to validate a model after it's built, never to fit it.

October 2026. Early access opens first to design partners — teams with a real question and real data who want simulation-backed answers before anyone else.

Yes. Every answer comes from a simulation you can open and inspect, model by model, down to the individual reaction.

No. Field pilots at McMaster University and the LifeCode Workshop 2026 put it in front of researchers with no modeling background, who built working mechanistic models from scratch.

A growing, validated library — E. coli, A. baumannii (wild-type + MDR), HIV-1, statin PBPK-PD, SARS-CoV-2, and more — so new work starts from the current state of the art.

Early access

Bring a real question

Design partners get in first — teams with a target, a strain, a compound, or a process bottleneck, and the data to check an answer against.