The buffer engine stores ("buffers") all incoming funds and lets its members withdraw according to a pre-defined ratio.
Every group has an address. Send funds to it, reference the address in another contract, and so forth. You can treat the group as a single entity. Think of it like setting up a company with multiple founders, but much more casual. Once a company is set up, the company acts as an independent entity, and the founders share the profits.
Pipe is versatile. Because it doesn't have a specific contract method you need to call, but is activated simply by sending money to it, you can use it in all kinds of contexts. You can use it as a minimal module in a DAO, as a recipient of an NFT contract royalty, as a way to split funds among NFT collaborators, and so on.
Pipe truly shines when it comes to composability. Because each pipe is immutable, you can trust that the split algorithm will never change in the future. Therefore can treat pipes as low level building blocks and build the addresses into oher contracts, and they will work deterministically forever.
Buffer is powered by merkle trees. Normally dealing with merkle trees is complex and requires offchain storage, and often results in some business running it as a centralized service. But not with Buffer. You DO NOT need to rely on a 3rd party. Here's how it works:
Therefore Buffer does not have any centralized cloud that stores your merkle trees, and has no 3rd party reliance. Everything is on IPFS and Ethereum. Anyone can run Buffer on their own.