Opening the Agent-Native Email Quest
Email is where too many relationship loops go to hide.
Not because email is bad. Email is still the common protocol for professional trust - the one channel every relationship in my life eventually touches. The problem is that an inbox is a stream pretending to be a system: opportunities, obligations, introductions, ideas, and promises all arrive in the same narrow chute, and the good ones sink.
I want a better shape for it. And I want to own the shape.
The Agent-Native Email project is my public build to fix both at once.
What I am building
Own the rails. My mail now lives on a small server I run myself, following Derek Sivers’ tech-independence checklist - my domain, my machine, open protocols, an exit that takes an hour. As of this week, my thrivinghenry.com mailbox receives on hardware I control. Dependencies are fine; hostage situations are not.
Teach agents to run them. The first hire on my virtual team - an executive assistant agent - already triages the inbox and drafts me a morning brief. It was tested against a golden set of labeled emails before it ever touched a real one, including a prompt-injection attempt it correctly flagged instead of obeyed. Nothing sends without my approval. Autonomy gets earned with evidence, class by class.
Publish the journey. Every piece of this build lands here: what worked, what broke (locking myself out of my own server: instructive), and what it costs - including the real token bills, which nobody publishes and I think they should.
The discipline is progressive disclosure: build the innermost working circle before drawing the larger map. Email first. If it works, it becomes the pattern for every channel after it.
What good looks like
The test is not whether the workflow feels clever. The test is whether fewer good things fall through the cracks - and whether I can trust the system enough that my old Gmail becomes a burner account where nothing important lands.
If someone makes an introduction, I follow through. If I promise a resource, it becomes a visible loop. If a relationship deserves care, the system helps me notice before neglect turns into apology. That is what I mean by agent-native: not replacing judgment - giving judgment better memory, better context, and better timing.
Live build notes stay on the quest hub, with an RSS means updates come to you - no account, no algorithm. Paste this into any feed reader: for anyone who wants to follow the work without joining another platform.