A cluster of announcements this week shows the AI agent identity stack splitting into distinct, separately governed layers rather than converging on one standard. 1Password published details on its integration with Anthropic's Claude that keeps credentials out of the model entirely. Once an agent can click, buy, update, and submit for a user, the company frames the core problem as figuring out what identity it's acting under and what access it should get. Its approach uses a zero-exposure architecture: Claude can complete browser logins, but "the credentials never enter the model or its memory," with 1Password injecting them directly into the page at runtime after biometric approval. 1Password's CTO Nancy Wang put the distinction plainly: Claude can know it used a login without ever holding the password or one-time code. Separately, Dai Nippon Printing announced it has begun offering digital identity management for AI agents making proxy purchases, built on its CATRINA distributed-ID platform and using verifiable credentials so that only agents authorized by a specific user can transact on that user's behalf. The most institutionally significant move is the operational launch of the x402 Foundation, a new open-governance body under the Linux Foundation, created by Coinbase to steward x402, an open standard for internet-native payments over HTTP. The protocol embeds payment capability directly into web interactions so agents, APIs, and applications can exchange payment the way they already exchange data, covering everything from card rails to stablecoins. The foundation already counts roughly 40 member organizations, including Amazon Web Services, American Express, Circle, Cloudflare, Google, Mastercard, Stripe, and Visa. Taken together, the pattern is a division of labor: DNP addresses an agent's authority to buy, 1Password addresses its authority to access the account needed to buy, and x402 addresses the transaction mechanics themselves. None of the three solve the whole problem alone, but each is narrowing what used to be a single, muddy question — is this agent allowed to do this, on whose behalf, using what credential — into separable, auditable layers. That separation is the more durable signal here than any one vendor's launch: it suggests the industry is settling on identity, authority, authentication, and payment as distinct planes rather than trying to bind them into one proprietary token.