So this product is a thin wrapper over a remote virtual machine, a new name for remote virtual machines, and an exhortation to not look inside or reason about remote virtual machines?
I feel like up until recently this would been pitched as “the missing UX/DX for cloud compute”, or something. But instead now it’s marketed to this new trend where one should aggressively not understand the tower of abstractions one is standing on. Odd.
I think very soon big LLM providers (OpenAI et. al.) will provide a service like this. With the LLMs becoming faster, the agentic task completion bottleneck will soon move to the tool calls (both execution time and round-trip latency), so it makes sense to have a server host the project very close to the actual LLM doing the inference in order to minimize latency.
I feel like up until recently this would been pitched as “the missing UX/DX for cloud compute”, or something. But instead now it’s marketed to this new trend where one should aggressively not understand the tower of abstractions one is standing on. Odd.