How it works
The intuition in three paragraphs.
At its core, MPC splits each input into shares that look random to anyone holding fewer than a threshold number of them. Parties exchange and combine these shares step-by-step according to a protocol designed to compute the function (addition, comparison, or anything reducible to those) and reveal only the final result.
Because no party ever holds more than its own share of any input, no party can reconstruct anyone else's data. Privacy is guaranteed by the math, not by trust or contract. The same property survives compromise of individual nodes up to the protocol's threshold.
Partisia's MPC implementations target the threshold-honest-majority regime used across regulated industries: finance, healthcare, public sector. [Placeholder for a deeper performance / threat-model explainer once the docs site catches up.]
Further reading