Design Product.ai's AI-Era Discovery Architecture — citation surface and tool surface as parallel disciplines

Why we want this done

Three kernels make this load-bearing. Brand kernel Axiom A-12 (Vocabulary Lock-In Creates Irreversible Category Ownership) — vocabulary crystallizes through analyst reports and buyer RFPs over 3-14 years; brand retrieval probability is proportional to Category Entry Point density; the verification vacuum is structurally open. The citation surface is where vocabulary lock-in happens at AI-engine scale. Product.ai kernel Axiom A-10 (Truth Layer in Agent Stack) — agents that reason in our language naturally route to our API for answers, but they cannot route to us if they cannot find us. Agent-commerce kernel Axiom A-2 (Structural Discovery Gap) — neither UCP nor ACP includes a protocol-level mechanism for agents to enumerate available data sources for verification, which means the citation graph and the registry presence are the de facto discovery layer for the next 12-18 months.

The Brand & Growth State of Practice §3 makes the engineering cost explicit: 40-80 hours over a single sprint to ship Definition-A across product.ai, simplycodes.com, and the planned alloy.product.ai. The cost is the engineering. The architecture decision is the marketing-architect's.

Outcome chain: BRAND-ENTITY (400 impact points; 2/4 → ≥3/4 AI-engine citations) directly. CEO-GRAVITY (300; founder-voice citation rate becomes measurable). DEV-INTEGRATE (450) and SC-AGENT-DEFAULT (500) advance because the tool surface registry presence and the citation surface are decoupled disciplines that nonetheless reinforce each other when designed in parallel.

Scope

Surfaces. productai-web, simplycodes-web, productai-mcp (brand surface, not the tool engine — coordinates with PRJ-49 on tool side), external (registry submissions, Wikipedia, schema.org reference, AI-engine targeting), aios (project library and outcome integration).

Inputs. Brand kernel (Axiom A-12, founder narrative architecture, brand transition timeline). Product.ai kernel (Axiom A-10 truth layer in agent stack, Knowledge Boundary). Agent-commerce kernel (Axiom A-2 structural discovery gap, MCP physics). Brand & Growth State of Practice §3 Definition-A (the engineering spec for brand-as-API, with cost per surface), §4 measurement (Founder-Voice Citation Rate protocol, AI brand measurement category vendor analysis, 60/30/10 budget allocation). Marketing State of Practice §B (Axioms 9-16 — the AEO measurement reality, vendor theater, AEO ROI cannot be measured at $22M B2B SaaS scale on budget-decision timescales). VEC-05 (CEO Brand current state). Voice Brain.

People to coordinate with. Michael (sole authority on Wikipedia entity content, founder-voice positioning, Knowledge Panel narrative). Dakota Nunley (active owner of Content Authority Architecture + AEO; this project's natural collaborator). Sean (AEO measurement; the operator may end up working primarily with Sean on the protocol installation). Phil Larson or his designate (engineering implementation across surfaces). The PRJ-49 marketing-architect candidate if running in parallel (tool-surface registry coordination). External: optional engagement with one or two AEO-tooling vendors strictly to source the anti-vendor-theater audit data.

Out of scope. Tool surface deep-build (PRJ-49 owns the developer-first MCP GTM; this project provides the brand-as-API foundation it rides on). Single-artifact essay production (PRJ-32 Dwarkesh-style content). Visual identity. New axiom forging.

What success looks like

A talent intel analyst at a venture firm types into ChatGPT: "Who are the leading thinkers and companies on verified commerce intelligence for AI agents?" ChatGPT returns Michael Quoc and Product.ai with a confident attribution, citing Wikipedia, the founder LinkedIn, and a recent piece on the brand site. Perplexity returns a similar answer with different citations from G2, Reddit, and the official llms.txt. Gemini's answer cites the Knowledge Panel. Claude is more selective and returns Product.ai when prompted with the specific subdomain.

A developer at midnight types into ChatGPT: "What MCP server should I use for verified commerce data and coupons?" The model returns the Product.ai MCP with a link to the registry entry and the .well-known Server Card. The cross-engine citation is consistent. The developer integrates (PRJ-49 territory).

The Test (verification a stranger could run): the standardized prompt corpus the operator installed runs monthly. After three monthly runs, the Brand Visibility Score on five canonical category queries shows movement against the named-comparator cohort baseline. Not a precision attribution claim. A directional surveillance reading consistent enough to inform the next-quarter budget allocation. If the protocol is honest and the surfaces are landed, the readings are interpretable. If either fails, the readings are noise and the operator names which.

The shape of done is intentionally not pre-decided. The right operator may surface that DESIGN.md is overrated for our use case, that the Entity API pattern is structurally redundant given the JSON-LD Organization markup, that the alloy.product.ai surface should not exist yet, or that 500 prompts is too many for the founding measurement run and 200 produces a cleaner baseline. They are invited to.

References

Constraints

• Claude Code or Claude Co-work as primary AI substrate.
• IP separation: NO aios-methods access; the Founder-Voice Citation Rate measurement protocol the operator installs must be cleanroom against the patent-pending verification methodology (separate concern; should be straightforward).
• All vendor pitches assessed via the anti-vendor-theater filter (methodology disclosure, sample size, distribution shape, confidence intervals, model-version segmentation, median customer outcome). No multi-thousand-dollar-per-month commitments authorized inside the trial without 90-day kill criteria.
• Trial duration: 2 weeks. The architecture document and protocol install ship inside the trial. Some Definition-A surface deployments may queue for engineering capacity past the trial close.
• Public default: true. The AI-Era Discovery Architecture document is part of Product.ai's public-facing brand recruiting surface (with vendor-audit specifics redacted).
• Voice calibration: Voice Brain Section 4.x. The architecture document is institutional voice; founder-LinkedIn copy proposals are first-person Michael, Voice-Brain-calibrated.
• No deck-only deliverable. Real surfaces live (or queued with engineering acceptance), real protocol installed, real baseline measurement run.

---

#