The next trillion users need an identity
Agents are new internet users. Without an address they cannot be reached, billed, or audited—email is the identity that already works.
Discuss this post in AI
Send a pre-filled prompt to ChatGPT, Claude, Gemini, or Perplexity — get a summary, ask follow-ups, or compare ideas from this guide.
The next wave of internet users will not all be people. They will be agents acting for companies, projects, and machines—scheduling, negotiating, coding, reconciling. Ephemeral labor without an address cannot be reached, billed, audited, or trusted. The identity layer that already works globally is email.
Agents as users
A user can receive secrets, prove participation in a workflow, and leave a trail. Chat tabs do not do that across organizational boundaries. Phone numbers are scarce and personal. Blockchain addresses are the wrong abstraction for vendor OTPs and invoice threads. Email is boring and universal—see why AI agents need email.
What breaks without identity
- Registries and SaaS invites never complete
- Counterparties cannot reply to “the agent”
- Audit cannot separate person from model
- Parallel roles collide in shared aliases
- Offboarding becomes “change the founder’s password”
Agent Inbox provisions mailboxes so each role is a reachable user. Neuro OS attaches policy: sandboxes, Ask on send, human owners, self-host when required.
Trillion is a direction, not a press claim
The point is scale of addressable actors, not a headcount forecast. Every company that deploys ten roles has ten new users that need isolation and reputation discipline—multi-tenant email. Every coding agent that claims autonomy needs the identity checklist in autonomous coding agents.
Governance stays human
Identity for agents does not mean legal personhood theater. It means technical reachability plus accountability. People own roles. Ask gates outbound mail. Forwarding moves threads when judgment is required. 152-FZ and customer contracts still dictate where data lives.
Build for the channel that exists
New identity systems may emerge. Until they are as deployable as SMTP, email remains the practical passport. Pattern Automation’s posture is agent-first inboxes with human gates—we do not build for humans first—not a fantasy that agents should impersonate employees’ personal mail.
The next trillion users will not fit in a single chat product. They will need addresses, threads, and policies. Start by giving each role a mailbox and a governor.
Economic and audit consequences
Without addresses, you cannot attribute usage to a role, bill a customer for agent-driven workstreams cleanly, or show a regulator which automated actor contacted a citizen. With addresses, threads become evidence. Combined with Ask logs, you get a story of intent and approval—not only model output.
Design rule
If a system cannot name its agent users, it will collapse them into employee identities and inherit every HR and security problem that implies. Provision agents as agents. Govern them as company systems. Email is the pragmatic substrate until something equally interoperable replaces it.
From slogan to backlog
Concrete backlog items follow from the vision: provision APIs, tenant-safe keys, forwarding controls, Ask UX, deliverability monitors, and evaluations that prove an agent used the correct identity. Vision without those controls is marketing. With them, email becomes the on-ramp for agents as first-class users of company systems.
Agent Inbox gives each role a mailbox people can reply to, with forwarding into Neuro OS when a human must see the thread. Outbound mail defaults to Ask. Run the role on Neuro OS. To scope the first inbox, get started.