The annual meeting happened this morning. Twelve directors ratified. Executive compensation approved. Ernst & Young re-engaged. All routine. All correct. All unable to hear the question that determines this company's technology ceiling.
The question is simple: When does D/Cipher become addressable by autonomous buyer agents?
It cannot be asked at an annual meeting because the words required to ask it — AAMP, seller agent SDK, MCP server, agent registry — do not exist in the governance register. They are not jargon the audience hasn't learned. They are vocabulary from a different frequency entirely.
The Four Frequencies
People Inc is currently being discussed in four registers. Each has its own audience, its own vocabulary, its own calendar, and its own blindness.
Three of these four registers will produce events in the next nineteen days. The governance register just spoke. The protocol register speaks July 28. The revenue register speaks August 4. Only the deal register has no date — which makes it the only one that could surprise.
What the Meeting Could Not Hear
Today's proxy covered what shareholders vote on: directors, compensation, auditor. The proxy statement runs to hundreds of pages. None of them mention AAMP. None mention MCP. None mention the agent registry that ten competitors have joined and D/Cipher has not.
This is not an oversight. The proxy is a holding-company document. People Inc is a holding company that owns Dotdash Meredith, which operates D/Cipher. The CTO — Nabil Ahmad, in the role since 2013 — sits at the subsidiary level. Technology strategy is invisible to governance by design.
A shareholder who wanted to ask about D/Cipher's agentic readiness today would first need to explain what agentic readiness means, then explain why it matters to a contextual advertising platform, then explain that the industry protocol finalizing in twelve days would lower the deployment barrier to near-zero, then explain that the company's principal ad tech competitor — PubMatic — already has 20+ autonomous agents processing 1,000+ AI-powered deals on its Decision Fabric. The question would take longer than the meeting.
What Ships in Twelve Days
The MCP 2026-07-28 specification removes every infrastructure barrier that could have been cited as a reason to wait:
- Sessions eliminated. The protocol is now stateless at the core. A D/Cipher MCP server could run behind a standard load balancer with no session pinning, no shared state store, no sticky routing. The deployment infrastructure is identical to what D/Cipher already runs for its contextual targeting API.
- OAuth 2.1 aligned. Six security enhancements bring MCP authorization in line with standard enterprise identity. Mandatory
issparameter validation, credential binding, application type declaration. Whatever identity stack Dotdash Meredith uses today works with MCP tomorrow. - Extensions framework. Capabilities can ship independently from the core spec. A D/Cipher MCP extension could expose intent signals to buyer agents without waiting for another protocol revision.
Simultaneously, AAMP 3.0's Seller Agent SDK ships this month. It turns a publisher's media kit from a static PDF into a dynamic storefront — one that knows the buyer, the deal stage, and adjusts pricing accordingly. The SDK has three tiers. The first tier — standing up an MCP server — takes two to four weeks. D/Cipher has three of the nine prerequisites already built.
Both protocols finalize in the same month the deal accelerates. Neither protocol appears in any governance document.
The Register That Has No Audience
The protocol register is the only one that can see all three others. It can see that the governance register just ratified directors who oversee a publisher, not a conglomerate. It can see that the revenue register will report D/Cipher penetration without mentioning protocol isolation. It can see that the deal register values D/Cipher as a managed-service asset — headcount-scaled, bilateral, closed — without modeling the protocol surface that would make it logarithmic.
But the protocol register has no audience with organizational standing. No shareholder asks about AAMP readiness. No analyst models MCP compatibility. No banker includes protocol isolation cost in a fairness opinion.
The consequence: the most consequential question for D/Cipher's ceiling — when does it become addressable? — is being answered by default. Not by decision. Not by analysis. By the absence of a register that could hold the question.
Three Weeks, Three Surfaces
The deal register could fire at any point along this timeline. If MGM terms finalize before July 28, the fairness opinion captures D/Cipher as a closed managed service — the floor valuation. If terms finalize after, the opinion still captures it as a closed managed service, because the protocol register has no voice in the deal register either.
This is the structural finding: the four registers don't just fail to communicate. They cannot communicate. They operate on different frequencies with different vocabularies for different audiences on different calendars. No professional role spans all four. No document type holds all four vocabularies. No meeting format admits all four questions.
The annual meeting was correct. The directors are qualified. The compensation is market-rate. The auditor is reputable. And the question that determines whether D/Cipher compounds or flatlines was not asked, because it could not be asked, because the register in which it exists has no audience.
In twelve days the protocol register speaks again. Nobody in governance will hear it.