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
Four things you can do the day you get in
Ask in plain language
Run the experiment yourself
Inspect the mechanism
A library you can build on
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.
| Dimension | An LLM answer | Omphalos |
|---|---|---|
| Where the answer comes from | A statistical pattern in a training set | A simulation built first-principles from biochemistry |
| When something's wrong | It can sound right — with no way to tell | Trace it to the exact reaction responsible |
| Your data | The model is fitted to it | Used to validate a model, never to fit it |
| Coverage | A new tool per use case | Molecule to organism, one platform |
What one model actually holds
reactions and processes in Virtual E. coli's genome-wide simulation
genes expressed genome-wide (K-12 MG1655)
sigma factors and two-component sensing systems modeled
How a model gets built
AI provides the hands and tools; the human expert is the control tower responsible for scientific judgment.
- 01
Draft
Generative AI assembles model components — reaction networks, parameters, compartments — from literature and open-source models.
- 02
Inspect
A domain expert reviews the logic, assumptions, and parameters directly in L++ before anything is accepted. Genuine authorship.
- 03
Simulate
The model runs. Plausible-sounding but structurally wrong reactions tend to surface here, visible in the output.
- 04
Calibrate
The model is tested against experimental or omics data it was never fitted to, and revised.
- 05
Deploy
The validated model becomes reusable infrastructure — ported, extended, or deployed into a new application.
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
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.
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.