Core Model Concept
The Organizational Canon
Right now your approved facts live scattered across PDFs, Notion pages, and people's memory — which is exactly why AI tools guess at them. A Canon is MsgStack's fix: the authoritative, structured truth for an organization, organized into queryable, version-locked domains that AI models can inspect programmatically instead of approximating.
Anatomy of the Canon Model
MsgStack parses and maps your organizational entities into a queryable data graph.
Canon Domain
A namespace or department-level bounding context (e.g. Developer Specs, HR Guidelines, SOC2 Security, GTM Messaging).
Canon Entry
A singular, atomic, structured assertion (a rule, policy statement, disclaimer, or capability fact) that is version-controlled and locked.
Canon Owner (SME)
The functional lead or department SME who is authorized to draft, modify, and approve entries within their domain.
Derived Outputs
Transient outputs generated by AI (code, release notes, battlecards, support emails) grounded in the approved canon. They are never the source of truth.
One Graph, Three Departments
Grounding Types by Department
Marketing, Engineering, and Legal don't write the same kind of truth — a positioning claim isn't shaped like an API contract, and neither is shaped like a legal disclaimer. MsgStack gives each department its own schema (Message House, Engineering Spec, Policy Shield) on the same underlying graph, so they stay connected instead of living in three disconnected tools.
Product Marketing Schemas
Same graph, same locking mechanism — just the field names product marketing actually uses.
The flagship schema — positioning, proof points, personas, and channel variants for product marketing.
{
"grounding_type": "message_house",
"section_type": "proof_point",
"content": "98% of zero-day threats stopped proactively.",
"content_tier": "tier_1_locked",
"personas": ["CISO", "VP Engineering"]
}From raw text to version-locked facts
Ingested text documents are parsed, structured into explicit schema entries, and mapped to specific department owners.
Before — Unstructured Document
=== Raw Spec: V2 API Ingestion === Endpoints for the new ingestion API must run on v2 paths. Rate limit is capped at 100 req/sec. We must enforce TLS 1.3 for compliance. Security lead says SOC2 audit requires logging IP addresses. ==================================
After — Governed Canon Entry
# Canon Domain: Developer Specs ## Entry: API Limits Owner: Engineering Lead Rule: Rate limit cap at 100 req/sec. Status [LOCKED]: "Active on all /v2/ paths." ## Entry: Transport Security Owner: Security SME Fact [LOCKED]: "TLS 1.3 enforcement active." Audit: "Enforce IP logging (SOC2 Section CC6.1)"
Propagation & Drift
When an authoritative spec or policy updates, downstream outputs shouldn't drift. MsgStack maps explicit dependency links (INFORMS / DEPENDS_ON) between entries — including across departments.
Concretely: a Policy Shield legal_disclaimer entry can DEPEND_ON an Engineering Spec api_contract claim — say, a rate-limit disclaimer that depends on the actual published limit. Change the limit in Engineering Spec, and the dependent legal entry is flagged outdated until Legal reviews and re-approves it. No disclaimer quietly goes stale just because nobody told Legal the API changed.
Drift alerts downstream owners instantly.
Non-approved drafts are excluded from active search.
Turbovec handles retrieval with zero external calls.
Say it once. Say it right, everywhere.
Ready to replace hallucination with deterministic organizational knowledge? Plug MsgStack's MCP server directly into your AI clients — or just ask it from Slack or Teams.