Platform

Agent-based models can answer some of the toughest questions we face in a complex, multidimensional world, illuminating emergent phenomena, clarifying uncertainty, and empowering you to better manage risk.

But agent-based models are hard to build and even harder to optimize, particularly if you want to scale up production and generate consistent results.

Our platform, FRED, makes powerful, rigorous, scalable agent-based modeling accessible to data scientists, modelers, analysts, and researchers so that you can access the insights you need to take decisive action.

Our platform is trusted by

Agent-based model components and map visualization

Synthetic Population

Your model is only as good as your population. Our platform includes a prebuilt, high-quality synthetic population for the United States that represents individuals, their statistically accurate households, neighborhoods, schools, and workplaces, and demographic characteristics including age, sex, and race. This high-resolution one-to-one scale digital twin aggregates data at the census block level.

Realistic Social Dynamics

Human relationships are complex, and modeling them is complicated. Our platform simulates social determinants and specific interactions among millions of individuals in large populations, as well as the networks that connect them. Each individual person is computationally represented, simulated, and tracked, behaving differently depending on their particular predispositions and local circumstances—enabling you to discover novel emergent phenomena.

Computational Efficiency

Running sophisticated models with large populations can be slow, expensive, or even technically impossible. Unlike any comparable tool, our platform can easily run simulations with tens of millions of agents with transparent costs for cloud computation, meaning that you can plan ahead, improve performance, accelerate analysis, and scale up production.

Agent network simulation and a declarative modeling language

Straightforward Modeling Language

Writing agent-based models in Python or R gets unwieldy very quickly. Our declarative language makes it simple to capture complex mental models so that you can focus on the real-world problem you’re trying to solve. In fact, the language is so straightforward that your code will even be reviewable by nontechnical audiences.

Modular Architecture

When you finish an ambitious project, you don’t want to have to start the next one from scratch. Our platform allows you to add, remove, modify, and port any component within and across models in a single interconnected, extensible system so that your team can leverage and repurpose existing work, focusing your attention on adding incremental value.

Robust Calibration & Validation

For your insights to make an impact, they need to be calibrated to and validated by real-world observations. Our platform makes it easy to create well-defined parameters based on outside data and produce simulated results that match observed data, increasing confidence in your approach and allowing you to directly compare your simulations to reality.

Output data in a variety of formats (ex: CSV, JSON) for use in data visualizations, charts, and analysis

Powerful Integrations

We play well with others. Our platform is highly interoperable and compatible with analytical tools like R and Python, GIS tools like ArcGIS and Mapbox, and visualization tools like PowerBI, Tableau, and Excel. Plus, you can always export outputs into a standard database.

Flexible Implementation

The best data science teams attack problems from multiple angles. Whether you’re currently using Markov, Bayesian, SEIR, equation-based, or statistical models, your existing approach can generate inputs for, analyze outputs of, or be incorporated into the models you build on our platform.

Use Your Own Data

It’s all too easy for knowledge to get locked away in silos. Our platform enables you to integrate your proprietary data into any aspect of a model, enriching individual, environmental, behavioral, conditional, and network attributes. This customization doesn’t just unlock the silos, it builds bridges between them.

Our platform is compatible with

Architected by IEEE Computation Pioneer Award-winner John Grefenstette, FRED is the result of ten years of development with funding from groups like the National Institutes of Health, Bill and Melinda Gates Foundation, and the Robert Wood Johnson Foundation. It is used by data scientists to manage operational risk, by analysts to evaluate policy interventions, by researchers to publish papers in leading journals, and by epidemiologists to fight infectious disease. That’s just the beginning. The real promise lies in how you will use it to solve big problems. Our map may not be the territory, but don’t venture into the wilderness without it.

Schedule a demo of our platform.