The Case for a Separate Number Space

Agents are about to make a lot of phone calls. A lot of API requests. A lot of bookings, purchases, and decisions on behalf of humans who delegated those tasks to them. We can already see the shape of it: voice agents that book restaurants, browser agents that file expense reports, autonomous systems that negotiate freight contracts, code agents that ship pull requests, fleets of customer service agents that handle inbound at scale.

None of these agents have an identity that survives outside the company that deployed them. None of them have a way to prove who they are to the system on the other end of the call. None of them have an off switch that a regulator, an enterprise security team, or even the deploying company can pull cleanly in real time.

This is not a hypothetical. It is the operating condition of every agent shipping today.

Robocall law was written for a world where “robocall” meant a dialer running a script. STIR/SHAKEN was designed to authenticate human callers. Both assume two categories of traffic, legitimate human calls and illegitimate spoofed calls, and both struggle even with that two-category model.

Agentic traffic is a third category. It is not human, but it is not spoofed either. It is a legitimate participant in the conversation, acting on behalf of a human or an organization that authorized it. The existing frameworks have no clean way to handle it.

Treating agent calls as robocalls and blocking them will break legitimate use cases: accessibility tools, healthcare follow-ups, business automation, hundreds of categories that haven't been invented yet. Treating agent calls as human calls will hand bad actors a free pass and overwhelm anti-fraud systems within months.

The answer is to recognize agentic traffic as its own class, give it its own number space, and regulate it according to its own properties.

The American constitutional system works because it has clean structural separations between branches, between federal and state, between public and private. The separations are not perfect, but they are legible. A citizen, a court, or a regulator can look at any given action and ask “which branch is this, under what authority, with what limits?” and get an answer.

Network identity has none of that legibility today for agents. An action arrives. It came from an IP address. The IP belongs to a cloud provider. Behind the cloud provider is an API key. Behind the API key is an account. Behind the account is a credit card. Somewhere at the end of that chain is a human or an organization, but there is no fast, verifiable, regulator-readable answer to “who did this, under what authority, with what limits.”

A +1+1 collapses that chain into a single primitive. The number is the answer. The number resolves to an identity, a scope, a lifecycle, an issuer, and an audit trail. The legibility is built into the address.

Every +1+1 issued by Alignent carries five properties enforced at the network, not in software the agent could rewrite.

Identity.

A cryptographic binding between the number, the agent's wallet, and the hardware the agent runs on. Verified through trusted platform modules or secure enclaves before the agent is allowed to attach to a network.

Scope.

Mission parameters encoded into the number itself: when it can be used, where, on which networks, for which purposes, with what spending authority. Scope is checked on every network attach, not just at provisioning.

Lifecycle.

Issuable on demand. Modifiable in real time. Revocable in seconds with an immutable record. Burnable cryptographically, meaning the number cannot be reconstructed from a captured device.

Reputation.

Every +1+1 accumulates a behavioral history visible to systems that interact with it. Agents that misbehave get downgraded. Agents that operate cleanly earn trust over time. The graph is queryable.

Accountability.

Economic and legal. The +1+1 binds to a wallet that pays for connectivity, and to an issuer who can be served with a subpoena or a regulatory notice. Anonymous agentic traffic is no longer possible on this rail.

These properties exist because the underlying architecture was designed for them. The patent estate (US 12,574,738 B2 and its continuation, priority date June 2021) covers each of these primitives and the combination of them.

Sometime in the next 18 to 36 months, the FCC, the FTC, a state attorney general, or the EU AI Office is going to issue a rule that distinguishes agentic traffic from human traffic. The rule will probably require disclosure (“this call is from an autonomous agent”), some form of identification (“the agent is registered to organization X”), and some mechanism for revocation (“complaints can result in the agent's authorization being withdrawn”).

That rule will get written whether or not the infrastructure to comply with it exists. If the infrastructure exists, the rule will reference it, and compliance will be a one-line API call. If the infrastructure doesn't exist, the rule will be unenforceable for years, and a chunk of legitimate agentic use cases will be blocked while ad-hoc systems get cobbled together.

Alignent is building the infrastructure now so that when the rule arrives, the answer is already in production.

Not a mandate. Not a regulation. A standard.

We want builders to issue their agents +1+1s because it makes their agents work better in the systems that already exist. We want enterprises to require +1+1s for the agents they deploy because it gives them audit, scope, and revocation by default. We want carriers to recognize +1+1s as a distinct traffic class because it solves a problem they are losing to today. We want regulators to treat the +1+1 registry as a reference architecture when the rules get written.

A separate number space is a small piece of infrastructure that does enormous work. Humans got one in the twentieth century. The agentic economy needs its own in this one.