dfmchn • Difference Machine

dfmchn (Difference Machine) is a behavioral stress-testing engine. It reads what a population actually says, at scale, and models how that population moves through the stages that lead to a decision: staying or leaving, buying or walking, holding or switching. Then it finds the cliff, the point where behavior tips, before it shows up in anyone's dashboard. Most analytics tell you what already happened. A churn number climbs after the customers are already gone. dfmchn works the other way around. It models the friction building underneath, the reversible kind that never reaches a metric until it is too late, and shows you where a population is close to the edge while you can still do something about it. It is domain agnostic. It learns the friction from the population's own words, not from our assumptions, so you can point it at customers today, an electorate next week, and your own workforce after that. Same engine, no retraining. New crowd, new physics, figured out from the language. And it is deterministic. dfmchn is built on graph theory and simulation, not a chatbot guessing at answers. No model drift, no hallucinations, just math you can trace. You can push the variables yourself and watch the population respond. dfmchn is a Austin-based start-up working at the intersection of graph models, simulation, and human decision boundaries. We are small, technical, and direct. We care about getting the math right and saying only what the data supports.