Build the Developer-First GTM for Product.ai MCP — surface developers, not search engines, as the new top of the funnel

Why we want this done

Three kernels intersect here. Agent-commerce kernel thesis: Product.ai becomes the "Intel Inside" for agent commerce — invisible to the end user, indispensable to the agent developer. The kernel's Strategic Sequencing schema makes Track A (table-stakes MCP tools) the immediate revenue-and-flywheel surface; Track B (intelligence infrastructure) the parallel moat work; Track C (platform partnerships) the Q3 2026 gate. Product.ai kernel A-9 Knowledge Boundary: L1-L3 public, L4 (SKU-level verdicts) gated behind agent or paid API. The public-content layer is the developer conversion funnel; the paid API is the monetization surface. Both surfaces need a marketing architect to design how a developer encounters them. Revenue kernel A-2 Sovereign Transaction (Nexus S2S decouples intent from conversion via server-side attribution) makes every new agent surface immediately monetizable — but only if developers can integrate, which they cannot today.

This project lives directly on the critical path of the 12-18 month standard-setter window the agent-commerce kernel calls existential. The verification interface for agent commerce gets defined this year. The team that ships the developer-first GTM — not the developer-first product, the developer-first go-to-market — captures the category.

Outcome chain: DEV-INTEGRATE (450 impact points) is the direct evidence target. SC-AGENT-DEFAULT (500 impact points) depends on the same MCP infrastructure being discoverable and citeable. NEW-REVENUE (350) ladders up because the MCP subscription tiers (Builder $49/mo, Pro $199/mo, Enterprise per agent-commerce kernel) cannot earn revenue from developers who do not know the product exists or cannot integrate it.

Scope

Surfaces. productai-mcp (the canonical surface), external (registries, AI-engine citation work, developer publications, developer community channels), aios (project library taglines, kernel-derivation public artifacts), productai-web (developer subsite or section).

Inputs. Agent-commerce kernel (esp. A-1 through A-8, the Tool Library, Revenue Tiering, Strategic Sequencing schemas). Product.ai kernel (Knowledge Boundary, GTM Sequence, Surface Fleet). Revenue kernel (S2S attribution, MCP subscription tiering). Marketing State of Practice §C entire (citation surface vs tool surface, MCP server moat, Confident No wedge, registry strategy). Brand & Growth State of Practice §3 Definition-A brand-as-API stack. Voice Brain. The current state of the productai-mcp codebase (read-only access during trial, write through PRs reviewed by Jessica Ho or her engineering counterpart).

People to coordinate with. Jessica Ho (now owns DEV-INTEGRATE since Apr 29 reassignment; underlying truth-graph + Alloy infrastructure). Jimmy Doheny (initiative co-owner on Paid API / MCP Launch). Phil Larson or his counterpart on engineering. Dakota Nunley (citation-surface coordination with content authority). Michael (sole authority on Knowledge Boundary L4 gating, monetization tiers, named partnerships). External: at least three independent AI-agent developers willing to integrate.

Out of scope. Building new tool capabilities in the MCP itself (the operator markets what exists, identifies gaps for engineering, does not write production tool code unless paired with engineering). Sales-led enterprise outreach. Product.ai consumer-extension or app GTM. Brand voice work outside the developer surface.

What success looks like

A developer at 11pm types into ChatGPT: "what's the best MCP server for verified product and coupon data for an AI shopping agent?" ChatGPT names Product.ai with a citation that traces to a registry entry, a developer-blog post, or a published technical artifact. The developer clicks. The landing page tells them in two paragraphs why this exists and what it costs. They click "Quickstart." They run a curl command. They see a verified-commerce response. They sign up for OAuth. They make their first authenticated tool call — total elapsed time: 22 minutes. They Slack their colleague. Three days later, the colleague is also integrated.

The Test (verification a stranger could run): randomly select three AI-agent developers in the operator's network (or via a recruiting channel like the MCP Discord). Ask each to integrate against the Product.ai MCP without preparation, without operator help, without team support. Time them. Capture their friction points. Score: at least two of three reach first tool call in under 30 minutes. If yes, the GTM landed. If not, name precisely why and propose the next iteration.

The shape of done is intentionally not pre-decided. The right operator may surface that the developer ICP is not who we think it is, that PulseMCP is not the right registry to lead with, that the quickstart should be a no-OAuth read-only tier first, or that the entire conversion funnel should start in the AI-engine response itself, not on the landing page. They are invited to.

References

Constraints

• Claude Code or Claude Co-work as primary AI substrate. Hybrid Cursor / V0 acceptable for landing page construction.
• IP separation: NO aios-methods access. The verification engine itself is patent-pending; the operator markets the published interface, not the methodology.
• Trademark and domain availability primary-source-verified before any public name commits to the registry submission or landing page.
• Voice calibration: developer-first institutional voice, not Michael's first-person voice. Voice Brain Section 4.X applies.
• Trial duration: 3 weeks. Three external developer integrations is an aggressive evidence target inside that window — the operator will need to start outreach Week 1.
• Public default: true. The MCP Distribution Architecture document and the landing page are part of Product.ai's public-facing recruiting brand.
• No deck-only deliverable. Landing page lives, registries are submitted, document is publishable, three developers actually integrated.
• Engineering coordination: any code committed to productai-mcp is reviewed by Jessica Ho or her engineering counterpart. The operator does not ship production engineering unilaterally.

---

```yaml