Filecoin (https://filecoin.io/); an incentive layer on top of IPFS to create a decentralized long-term object storage system.
Tradelens (https://www.tradelens.com/); in development, AP Moller Maersk's enterprise container logistics/coordination blockchain.
Anytime someone says "blockchains don't do anything which centralized systems can't do", I mentally interpret that as someone in 1995 saying "email can't do anything fax machines can't do". In the sense that: its kind of true, in that they accomplish a similar goal of sending a document from one person to another over wire, but it discounts as irrelevant the most critical, foundational thrust of why this new technology is interesting. For email: that it is all-digital. For blockchain: that it is decentralized.
In the 90s, the people who still clung to fax machines didn't understand why an all-digital future was important or would matter. They had spent their entire lives living in physicality, working with paper and file cabinets, and it was fine. The fax machine made sense. Email didn't.
Sure, centralized systems can do many of these things better. But they do so with the sacrifice of being centralized! Some critics inexplicably gloss over this like its an irrelevant, minor part of the argument. They've spent their entire lives among big tech billion dollar centralized multinational conglomerates, and its been fine. More centralization makes sense; email, I mean, decentralization, doesn't.
The point is not to make a better system, in every way; the point is making a functional system that is decentralized, so it can operate in a trustless, geo-distributed, multi-party way.