How many lab runs does Augur need to make predictions?

As few as 5 lab runs. Augur uses Physics-Informed Neural Networks (PINNs) that embed conservation laws directly into the model architecture, enabling accurate production-scale predictions from minimal experimental data — unlike pure ML approaches that need hundreds of runs.

What organisms does Augur support?

Augur supports 6 organisms out of the box: CHO (Chinese Hamster Ovary), E. coli, Saccharomyces cerevisiae (yeast), Pichia pastoris, Pseudomonas putida, and custom organism configurations. Each organism has pre-configured kinetics, economics, and DSP templates.

How does Augur predict COGS/kg?

Augur integrates upstream bioreactor simulation (titer, rate, yield predictions) with full downstream processing cost modeling across 9 unit operations. It calculates COGS/kg including media, labor, equipment depreciation, utilities, and consumables — with Monte Carlo confidence intervals to quantify uncertainty.

How fast are Augur's predictions?

Predictions complete in under 30 seconds. Augur uses reduced-order models and an ensemble of 5 PINNs to deliver production-scale economics predictions in seconds rather than the hours required by traditional CFD simulation.

What is a physics-informed neural network (PINN) for bioprocess simulation?

A physics-informed neural network (PINN) is a machine learning model that incorporates physical laws — such as mass balance, energy balance, and transport equations — directly into its loss function and architecture. In bioprocess simulation, PINNs constrain predictions to obey Monod kinetics, product inhibition, and overflow metabolism, enabling accurate scale-up predictions from as few as 5 lab runs. Unlike traditional CFD which takes hours, or pure ML which needs hundreds of runs, PINNs combine the data efficiency of physics with the flexibility of neural networks.

How does Augur compare to traditional bioprocess simulation tools like SuperPro Designer or BioSolve?

Traditional tools like SuperPro Designer and BioSolve are powerful for steady-state process modeling but require manual parameter specification and extensive experimental data. Augur differs in three key ways: (1) it uses physics-informed AI to predict scale-up behavior from minimal lab data rather than requiring manual inputs, (2) it couples upstream bioreactor simulation with downstream processing economics in a single workflow, and (3) it includes uncertainty quantification via conformal prediction — so you know how confident to be in each prediction.

Can Augur predict fermentation scale-up from lab to production scale?

Yes. Augur predicts production-scale performance (1,000L to 200,000L) from bench-scale lab data (1-10L). The platform models hydrodynamic changes during scale-up — including kLa (volumetric mass transfer coefficient), mixing time, tip speed, and dissolved oxygen profiles — and couples these with cellular kinetics to predict titer, rate, yield, and COGS/kg at production scale.

What downstream processing operations does Augur model?

Augur models 9 downstream processing (DSP) unit operations: centrifugation, depth/tangential flow filtration, chromatography (Protein A, IEX, HIC), ultrafiltration/diafiltration (UF/DF), liquid-liquid extraction, precipitation, crystallization, evaporation, and foam fractionation. Foam fractionation is uniquely suited to biosurfactant recovery, exploiting the surface activity of glycolipids (rhamnolipids, sophorolipids) for solvent-free product concentration. Organism-specific DSP templates are provided for monoclonal antibody (mAb) production, E. coli inclusion body processing, yeast secreted protein, small molecule, and biosurfactant workflows.

Can Augur model biosurfactant production and foam fractionation?

Yes. Augur is the only bioprocess simulation platform with built-in foam fractionation modeling for biosurfactant recovery. The platform models glycolipid production (rhamnolipids, sophorolipids, mannosylerythritol lipids) in organisms like Pseudomonas putida and Starmerella bombicola. Foam fractionation exploits the surface-active nature of biosurfactants to concentrate product in foam — a solvent-free, low-energy alternative to liquid-liquid extraction. Augur lets you compare DSP strategies (foam fractionation vs solvent extraction) and predict COGS/kg to determine cost competitiveness against petrochemical surfactants.

How does Augur help biosurfactant producers compete with petrochemical surfactants?

Biosurfactants must achieve COGS below $1-5/kg to compete with petrochemical alternatives like fatty acid ethoxylates and linear alkylbenzene sulfonates. Augur predicts production-scale COGS/kg from lab data with Monte Carlo confidence intervals, showing whether your process can hit cost targets before you invest in production infrastructure. Break-even analysis identifies the minimum selling price, while scenario comparison lets you evaluate different organisms, feedstocks, scales, and DSP strategies. The platform includes organism-specific profiles for P. putida (rhamnolipids) with tailored kinetics, economics, and DSP templates.

What is bioprocess digital twin technology?

A bioprocess digital twin is a computational model that mirrors a physical bioreactor and its associated processing steps. It combines real-time or historical sensor data with physics-based simulation to predict process outcomes. Augur's digital twin approach uses physics-informed neural networks (PINNs) coupled with ODE kinetics modeling and integrated process economics to create a predictive digital twin from as few as 5 lab-scale fermentation runs — enabling virtual experimentation and de-risking before committing to expensive production-scale campaigns.

How does Augur handle uncertainty in bioprocess predictions?

Augur uses three layers of uncertainty quantification: (1) an ensemble of 5 PINNs with bootstrap sampling to capture model uncertainty, (2) conformal prediction calibration using held-out lab data to produce statistically valid confidence intervals, and (3) Monte Carlo simulation for COGS/kg economics to propagate uncertainty through cost calculations. Risk factors are automatically identified and ranked by severity, including out-of-distribution detection for extreme scale-up ratios or process conditions outside the training data range.

Augur — Bioprocess Intelligence Platform

Predict production-scale bioprocess economics from as few as 5 lab runs using physics-informed AI. Augur delivers COGS/kg, titer, rate, and yield predictions in seconds.

What is Augur?

Augur is a bioprocess simulation platform built on physics-informed neural networks (PINNs). It predicts production-scale fermentation performance and cost of goods (COGS/kg) from minimal lab data — as few as 5 fermentation runs. Unlike traditional CFD simulation that takes hours, or pure machine learning that needs hundreds of runs, Augur combines the data efficiency of physics with the flexibility of neural networks.

Key Capabilities

Physics-informed scale-up prediction (lab to 200,000L production)
Integrated COGS/kg economics with Monte Carlo confidence intervals
8 downstream processing unit operations with organism-specific templates
Support for CHO, E. coli, S. cerevisiae, Pichia pastoris, P. putida
AI co-pilot for bioprocess optimization and root cause analysis
Conformal prediction for statistically valid uncertainty quantification
Break-even analysis and sensitivity tornado charts

How It Works

Upload CSV data from your lab-scale fermentation runs
Configure production scale parameters (volume, organism, DSP train)
Get production-scale COGS/kg predictions with confidence intervals in under 30 seconds

Request Early Access | Schedule a Demo