User's guide

Scillion is a system dynamics modeling platform designed to help you understand and predict how complex systems evolve over time. Whether you're managing a business, forecasting market behavior, or analyzing organizational dynamics, this tool provides a structured approach to capturing cause-and-effect relationships and simulating their outcomes.

The platform addresses a fundamental challenge: real-world systems are interconnected. Changes in one area ripple through others in ways that aren't always obvious. Traditional spreadsheets make it difficult to model these interdependencies. Scillion solves this by providing a visual, equation-based modeling framework where you define how different elements of your system interact, then watch how they evolve through time.

At its core, scillion follows the system dynamics methodology, a proven approach used by researchers, strategists, and planners to understand complex problems in business, economics, public policy, and environmental science. You'll build models by defining stocks (things that accumulate), flows (what causes change), and parameters (the rules that govern behavior), then run simulations to see what happens.

Scillion looks forward to operating on the next principles: transparency and interpretability. Every relationship in your model is explicit and visible. The rules are defined, equations are visible, and everything can be understandable to clarify why your system behaves the way it does. Strategic decisions require justifiable assumptions.

The engine powering scillion is built on differentiable equations. Every relationship in your model is structured mathematically, enabling advanced computational capabilities. While classical system dynamics models are powerful for understanding behavior, they're often hard to optimize. Scillion adresses this matter by making its models autodifferentiable. This means we can automatically compute sensitivity gradients, showing exactly how changing each assumption propagates through your entire system. Want to know which relations push forward some output? The engine computes it instantly. This combines the clarity and interpretability of classical system dynamics with modern AI optimization techniques, achieving both understanding and deep-tech power.

This approach is particularly valuable when you have domain knowledge but limited historical data. You might not have years of customer transaction data for deep learning to work with, but you understand your business deeply. You know your sales cycle, your cost structure, your team dynamics. Scillion lets you capture that knowledge as explicit models, then simulate and test out different outcomes, playing around with confidence because every assumption is visible and justifiable.

Moreover, you can learn patterns from your historical TIME-SERIES FORMAT data, using it to estimate parameters or validate behavior. This hybrid approach combines the best of both worlds: you can build models based on your understanding of the system and/or then refine them with data to improve accuracy and confidence, or viceversa.