There are two main differences between custodial and non-custodial: A) Insurance. Which is a feature that can be less relevant when speaking about stablecoins, which are more like MMFs and therefore narrow banks, where the credit transformation risk is far lower. Moreover, it is also a feature that could be added over time thanks to capital markets. B) Compliance responsibility. If you offer self-custody accounts, you don't need to be a financial institution. However, if you want your customers' accounts to talk to traditional rails, you are only shifting the compliance effort to another part of the chain - for example to Bridge or Iron if you are connecting bank accounts, or to acquirers and issuers if you are connecting them with cards. At the end of the day, imo, self-custody will win, simply because it is much easier and faster to build a neobank by outsourcing pieces of compliance to external providers and orchestrating all of them to build the best ux for your customers. It is only a matter of which method is the most effective for building successful fintech businesses.

The vibes in the agentic commerce market are that things are both so early (no volumes) and so late (if you want to make a company here is already super crowded). I think this is because the payment primitives are easy to replicate, like crypto neobanking generally. The real game changer here is being able to identify customers with a problem where the agentic payments are the solution. And no one has been able to do it so far. I think there will be a lot of problems that can be solved by agentic payment neobanks, the fact is that they are arisen only for a ultra niche market so far. All of this means less attention from incumbents and fintech players, given the small volumes transacted and the shift in value networks that frictionless agentic payments create.