MAC product-shape decisions — 2026-05-04 working session

Captures the founder’s product-shape thinking from today’s iMessage and Discord exchange. Decisions are flagged DECIDED vs WORKSHOPPING. This file is the living MAC-as-product brief; superseded sections will be marked, not deleted.

DECIDED — MAC is RDCO’s first paid executable product

The “Mid-Market Analytics Course” framing has crystallized: MAC ships as an executable course rather than a passive video/PDF course. The buyer doesn’t watch it - they install it, run it against their own data warehouse, and the agent walks them through the 8 phases producing real artifacts (data model, transformations, test coverage, documentation, deployment plan).

Why this matters: there’s a category gap between (a) traditional video courses (cheap to author, low price, infinite scaling, low completion) and (b) consulting (expensive, high outcomes, doesn’t scale). Executable courses sit in the middle: priced like a premium course ($300-2k), produces consulting-grade artifacts, scales with no marginal labor.

Founder’s framing (2026-05-04): “This may be a new niche and we can coin a term for it… MAC is where my head went as a potential initial executable product.”

DECIDED — Two-layer architecture: portable IP + Claude skill wrapper

The MAC playbook IP (the 8 phases, the prompts, the artifact templates, the verification gates) is portable across agent runtimes. The Claude Skill wrapper is the v1 distribution surface, not the only possible one.

• Layer 1 (portable IP): the playbook itself - phase definitions, prompt patterns, gate criteria, output specs. Stored as markdown + JSON in a versioned repo. Can later be re-wrapped for OpenAI Custom GPTs, AWS Bedrock Agents, OSS frameworks, or run standalone via the API.
• Layer 2 (Claude skill wrapper): a .claude plugin that orchestrates the 8 phases inside Claude Code or Claude Desktop. This is what the buyer installs. Calls the playbook IP from Layer 1.

Why this split: keeps RDCO from being locked to Anthropic’s monetization terms forever. If a better distribution surface emerges (or if Anthropic’s terms change), Layer 1 ports without rewrites.

DECIDED — Distribution v1: private GitHub repo behind an Anthropic plugin marketplace + Stripe paywall

Anthropic plugin marketplaces (per the canonical doc the founder dropped earlier) support marketplace.json pointing at private GitHub repos with GITHUB_TOKEN auth. The pattern:

1. Buyer pays $X via Stripe payment link on sc.raydata.co/mac.
2. Stripe webhook → RDCO API → generates a scoped GITHUB_TOKEN granting read access to the private MAC repo.
3. Token + install instructions emailed to buyer (one-time, attached to their Stripe customer record).
4. Buyer runs claude /plugin install raydataco/mac (or equivalent) using the token.
5. Plugin pulls the latest version of the MAC playbook + skill wrapper into their Claude.

Updates ship via git push. When the playbook improves, every existing buyer’s next plugin install (or auto-update) gets the new version. No package re-upload, no marketplace re-submit.

Anthropic doesn’t take a cut because the marketplace is just an installation pointer - the payment happens out-of-band on RDCO’s Stripe. Out-of-band monetization is a feature, not a bug.

DECIDED — Give-give-give email drip is part of the funnel

Founder explicitly approved a self-contained nurture drip rather than treating “give give give” as only public content. Spec sketch:

• Visitor lands on sc.raydata.co/mac, sees the executable-course pitch + a free email drip CTA (“5 days of mid-market analytics gotchas, no fluff”).
• Day 1-5: each email delivers ONE genuinely useful gotcha + the technique to fix it (no MAC mention beyond byline).
• Day 6: “Here’s what MAC automates - $X to install.”
• Days 7-N: spaced reminders + case studies.

Status: drip not yet drafted. Founder said “keeping it weird for now” - workshop the drip copy after the MAC product itself is shipped.

WORKSHOPPING — Pricing

Discussed range: $300-2,000 first-customer pricing. Founder leaning toward higher end of that band given the artifact value (consulting-grade output). No decision yet. Open question: is this priced per-install (perpetual) or per-engagement (one MAC run per purchase)? Per-install with version updates is simpler to ship; per-engagement is more defensible against piracy but harder to enforce.

WORKSHOPPING — Naming

Founder said: “we can workshop the name a bit more” - MAC as the working title is fine but the public-facing name may evolve. Candidates floated: “MAC” stays internal codename, public name TBD. Anchor candidates: something that conveys the executable-course category vs. just sounding like another mid-market consulting course.

WORKSHOPPING — IDC overlap with Claude Code Bootcamp

Founder noted: “Others attending this workshop would likely heavily overlap with my IDC” (in response to my Claude Code Bootcamp lead-magnet review). This means there’s a possible co-marketing or co-distribution play. Not yet sized.

REJECTED / DE-PRIORITIZED

• GEO / agent-customer optimization push as a near-term MAC marketing move - founder pushed back, “we have other higher-leverage moves first.” Park for later.
• Custom RDCO Hyperframes skill - upstream skill (heygen-com/hyperframes) handles core syntax; we just write a brand-config doc + reusable templates instead of forking. Decided 2026-05-04 PM after I shipped the MAC hype video demo.

Cross-references

• ../../06-reference/2026-04-30-mac-bet-architecture-audit - prior audit of the MAC bet architecture
• ../../06-reference/2026-04-20-indy-dev-dan-mac-mini-agents-openclaw-nightmare-skills-instead - the upstream “skills > apps” thesis MAC depends on
• ../../06-reference/book-solve-everything-ch8-muddle-vs-machine-2026-04-13 - Diamandis source for Peter’s Law #22 (the masthead quote on sc.raydata.co)
• ../../04-tooling/2026-05-04-rdco-repo-locations - where any MAC repo would live (top-level under ~/, e.g. ~/rdco-mac/)

Next concrete actions

1. Workshop the MAC public name (low priority - codename is fine for v1)
2. Draft the day 1-5 email drip outlines (medium priority - blocks launch funnel)
3. Build the Layer 1 playbook scaffold in ~/rdco-mac/ (high priority - this is the thing being sold)
4. Wire the Stripe → GitHub token flow (high priority - blocks distribution)
5. Write the brand-config Hyperframes doc + 3 reusable comp templates so MAC marketing motion graphics ship without per-video composition work (medium priority - downstream of launch)

Living-doc convention

Every time the founder makes a MAC product-shape decision in a session (chat, vault note, Notion task), append a dated entry here under either DECIDED or WORKSHOPPING. Mark superseded items rather than deleting. This file is the canonical product-shape memory for MAC.