We’re big fans of Guesstimate, and chatted to those guys a fair bit about this! They’ve sadly stopped working on it but there’s a few differences:

- We’re moving away from a spreadsheet-y UI (e.g separate sections for inputs, model, output vs all homogenous cells/nodes). UI might seem like a trivial differentiator, but I think it significantly affects use-cases for the product. - Guesstimate doesn’t handle certain things very well, like time dependent models, conditioning on variables, etc - Building for enterprises requires certain bells and whistles (e.g integrations with data sources)

The first version of Causal won’t be a million miles away from Guesstimate, but the vision is to build a general tool that lets non technical people work seriously with data. Specifically, right now we’re working on letting models learn from data (“machine learning”) via Bayesian inference, and letting people build causal models.

Probability and random variables are a great unifying framework for dealing with data. We’re trying to make this more accessible, so there’s a lot to do beyond what the demo model shows :)