Some time over the Christmas holidays, I experienced what I called a moment of "existential clarity" about AI and it's ramifications – when I realized that in the not-so-distant future, it was entirely possible that most of easyDNS' customers would be autonomous AI-driven agents rather than people.

Our internal project to completely rebuild our UX (still ongoing) was close to a quarter in, and it occurred to me that we could be building a bridge-to-nowhere. Why are we creating more elegant ways to render forms that input hostnames and their respective rdata when:

• you could probably just tell the backend what you want for your domain functionality to be and it can generate the requisite zonefile to facilitate it, and then

• not long after that every API is going to sit behind an MCP server and it'll all be done agentically via automated endpoints anyway.

What was the point? This question still bothers me, but we continue to toil away at the UX rebuild, because even though this is where everything is headed, there will still be a temporally long-tail of copy-pasting IP addresses into forms (in the meantime I spend my spare time vibe coding alternative ways to convey DNS and metadata to a zonefile rendering engine. I can see why this isn't totally a thing yet, but it will be.)

Recently, I started reading John W. Munsell's "Ingrain AI" – it hits the ground running, with the introduction titled "Every CEO's Nightmare", wherein it lays out the "productivity" induced death-spiral many companies may be blundering into, should they be pursuing AI merely as a cheat-code toward hyper-efficiency.