Experiment: research pack → site (Boring Company)

Table of contents

What this experiment is

This is not a polished magazine feature. It is a first deliberate pipe-through of a structured GROK Research pack onto NiftyChatGPT — filed under the Grok hub because:

1. Grok Build ran the research pipeline (transcript, claims, Exa expansion, synthesis).
2. The subject is another Musk-ecosystem company often discussed next to Tesla / xAI tooling culture.
3. We want a public place to see what the pack becomes as site content — so we can improve the pack → blog handoff later.

Pack path (local): GROK_Research/packs/boring-company-intro
Site series: tag boring-company + model grok

Pack → site mapping (v1)

We do not paste the entire Exa dump or claim files into the CMS. Synthesis first; raw stays in the pack.

Posts in this experiment

1. What The Boring Company actually ships (2026 field notes) — main readable article
2. Open questions after the Vegas tour pack — research backlog as public notes
3. This page — process meta

Rules we followed

• Label confidence: design targets ≠ measured facts.
• Surface conflicts: e.g. official pages disagree on passenger/station counts.
• Attribute sources: company site, on-tour tour video, secondary Exa leads.
• No legal advice: easement / “clipping” questions stay as questions.

What to improve next (process)

• Auto-generate a “claim table” component from pack YAML.
• One-click “publish synthesis stubs” script pack → src/content/posts/research/.
• Pull only Exa URLs with quality ≥ threshold into footnotes.
• Optional: map assets once we have a stable public Vegas Loop map.

Bottom line

If this series feels rough, that is intentional. The goal is a visible loop: research pack → Astro posts → human critique → better pipeline.