| name | business-content-core |
| description | Shared reference for the business-content cluster: the source-derived Voice Profile every lane reuses, the research/evidence substrate, the per-channel message contract, and the publish/market/fund/reach lane map. USE WHEN drafting any launch-, reputation-, or fundraising-sensitive content, building a Voice Profile, or planning multi-channel outbound — the interlocking conventions every spoke shares. |
| cluster | business-content |
| version | 1.0.0 |
Business-Content Core
Shared model for the business-content cluster. The publish, market, fund, and reach spokes
all depend on the same two upstream artifacts and the same channel discipline — keep them
consistent here so no spoke contradicts another or drifts into generic AI copy.
1. The model the cluster turns on: voice-first, evidence-backed, channel-native
Every output in this cluster is the same pipeline, regardless of lane:
Real sources ──derive──> VOICE PROFILE ─┐
├──shape──> Draft ──fit──> Channel-native output
Market/competitor facts ──ground──> Evidence ─┘
Two upstream artifacts feed everything downstream:
- VOICE PROFILE — a reusable style fingerprint built from real source material (posts, essays, launch notes, docs, site copy), not platform exemplars. It is the single source of truth for tone; downstream lanes consume it, they do not re-derive style per draft. →
brand-voice
- Evidence — market size, competitor positioning, fund/diligence facts with source attribution. Business claims trace to this, never to invention. →
market-research
Rule: if an output is voice-sensitive, the VOICE PROFILE is an input, not an afterthought; if it makes a business claim, that claim is sourced. Missing either is a defect, not a style choice.
2. The four lanes (and how they compose)
- Publish (own channels) →
article-writing (long-form), content-engine (platform-native systems, repurposing), seo (search visibility).
- Market (the launch) →
marketing-campaign — itself an orchestration layer that pulls brand-voice, content-engine, and seo into one multi-channel launch.
- Fund (investor-facing) →
investor-materials (decks/memos/models, kept mutually consistent), investor-outreach (cold/warm/follow-up/update emails).
- Reach (outbound & network) →
lead-intelligence (full find→score→path→enrich→draft pipeline), social-graph-ranker (the standalone bridge-ranking engine it runs internally).
Lanes share the substrate, so a single launch can cross several: brand-voice → marketing-campaign → content-engine/seo, then lead-intelligence for outbound — all on one VOICE PROFILE.
3. The channel contract (why "reuse the copy everywhere" is a tell)
A message is shaped for its surface; the same paragraph pasted across email, LinkedIn, and X is
an anti-pattern. Pick one primary channel, in warmth order:
| Channel | Use for | Shape |
|---|
| Warm intro (email) | highest-value asks with a viable bridge | one ask, one concrete reason, forwardable blurb |
| Direct email | cold high-value, investor, partnership | plain specific subject, open from something real |
| LinkedIn DM | target active there / stronger graph there | shorter than email, no fake professional warmth |
| X DM or reply | high-context operator/builder/investor | tightest; reference something real from their timeline |
Channel selection and the bridge math (which targets earn a warm intro vs. cold outreach) come
from social-graph-ranker; the drafting voice always comes from the VOICE PROFILE.
4. Conventions
- Voice before draft. Run
brand-voice first for anything launch-, reputation-, or fundraising-sensitive; reuse the latest confirmed VOICE PROFILE across related tasks in the same session.
- Source before claim. Route business/market claims through
market-research; attribute them.
- Consistency across a document set. Investor assets (deck, one-pager, memo, model) must agree on the same numbers and narrative —
investor-materials owns that consistency.
- Repurpose, don't re-author.
content-engine adapts one source asset across platforms without flattening the author's voice.
- No fabricated metrics, logos, or traction. If a number isn't sourced or supplied, mark it as an assumption.
5. Shared guardrails
- Voice-first: one
VOICE PROFILE flows everywhere; never default to generic AI cadence or re-derive style per draft.
- Evidence-backed: every business claim traces to a real source; flag assumptions explicitly.
- Channel-native: one primary channel, message shaped for it; identical copy across channels is a tell.
- Draft, don't send: produce outreach as drafts; never auto-send without explicit user approval.
- Ban the slop: kill curiosity-bait hooks, "not X, just Y", "no fluff", "game-changer", "excited to share", forced lowercase, founder-journey filler, social-proof stacking, and bulk-merge field tells.
- Which spoke: the engine (
social-graph-ranker) when the user wants the ranking/math standalone; the pipeline (lead-intelligence) when they want the full outbound workflow.
