A wallet address tells you an account exists. It doesn’t tell you who or what controls it, what it’s authorized to do, or whether it’s the same actor you dealt with last time. Agents operating without more than that run into a specific, recurring problem.
A pseudonymous address is a reasonable identity primitive for a human who initiates a transaction, waits for confirmation, and moves on. It becomes a weaker fit the moment the entity behind that address is an autonomous agent completing a multi-step task across several interactions, potentially with counterparties it has never dealt with before and may deal with again. The specific gap is this: an address proves that a key signed something. It doesn’t establish who or what is behind that key, what that entity is authorized to do, or whether the entity a counterparty is dealing with now is the same one it dealt with in a prior interaction.
What breaks without persistent identity
Consider an agent that needs to verify a counterparty before executing a task, locate a service it hasn’t used before, and settle a resulting payment. Each of those steps depends on some notion of identity. Verifying a counterparty means confirming that the entity on the other end is who or what it claims to be. Locating a service means trusting that the service registered under a given name is actually operated by the entity it claims to be operated by. Settling a payment to a recipient means confirming the recipient identity is the one actually entitled to receive it. An address alone answers none of these questions with any confidence, because an address can be generated freely, with no persistent connection to any claim about who controls it.
Permissions have the same problem. If an agent is authorized to spend up to a certain limit, interact with specific contracts, or act within a given scope, that authorization has to attach to something that persists across the agent’s actions. An address that could belong to anyone, with no verifiable history, offers nothing for a permission system to anchor to beyond the keys themselves — which is a weaker guarantee than permissions tied to an identity that was established once and verified consistently afterward.
What persistent identity actually provides
PPAL, Lithosphere’s identity layer built on the LEP100 standard, is designed around establishing identity once and carrying it forward through subsequent interactions, rather than treating each interaction as a fresh, unconnected event. An agent verified through PPAL carries that verification into execution, into service discovery, and into settlement, without re-proving who it is at each step. This is what makes reputation meaningful in an agent context: a reputation only has value if it’s attached to something that persists across multiple interactions, rather than resettable the instant a new address is generated.
DNNS complements this by handling discovery — letting agents and services find each other reliably within the same architecture that manages identity, rather than through an external lookup system with no connection to the identity layer. The two are related but distinct problems: PPAL answers who an agent is; DNNS answers how it’s found and reached. An agent that can prove its identity but can’t reliably locate a counterparty is still stuck. An agent that can locate services but has no verifiable identity to present once it finds them is equally stuck.
Why this becomes more important, not less, as agents proliferate
The practical cost of weak identity scales with the number of agents and the frequency of their interactions. A single agent completing occasional tasks can get by on relatively thin identity guarantees, because the cost of a bad interaction is contained. A network with many agents conducting frequent, multi-step interactions with counterparties they haven’t necessarily dealt with before needs identity to be a reliable, persistent property — not something reconstructed from scratch, with no history behind it, at the start of every interaction.
The underlying point is narrower than “AI needs blockchain,” which is a framing broad enough to mean almost nothing. The specific claim is that autonomous agents operating across multiple systems, verifying counterparties, and building any kind of track record need an identity layer that persists and is verifiable — not just an address that happens to have signed something. PPAL and DNNS exist because that gap is a real, specific, and increasingly common problem for agent-driven activity, not because identity is a generically good feature to add.
Source: https://lithosphere.network/why-an-autonomous-agent-needs-an-identity-not-just-an-address/