What is actually in one
A single deploy call materialises the whole specification in dependency order, reference data, then the dimensional model, then ontology types, skills, agent templates, workflow templates, the edge profile and the compliance mappings, with per-cell version tracking, so a pack can be rolled back, upgraded, or run side by side with its previous version.
- 01
A dimensional model
Facts, dimensions, slowly-changing-dimension rules, an industry calendar, metrics and compliance-tagged measures, over six conformed dimensions shared across industries.
- 02
Ontology types
Ten to fifteen object, link and action types registered in the knowledge graph, the domain's nouns and verbs, typed.
- 03
A skills library
Ten to twenty domain skills across eight skill types, callable by an agent and publishable from an existing catalog asset.
- 04
Agent templates
Three to five templates with pre-wired skills, data-access policy and escalation protocols.
- 05
Workflow templates
Five to ten executable graphs with human approval gates on the consequential steps.
- 06
An edge profile
Collectors, telemetry schemas, authorization policies, and extraction rules that map raw protocol fields to the object model at the point of capture.
- 07
Compliance mappings
The industry's regulations woven through every layer above rather than bolted on beside them.
Plus reference-data bundles: cross-industry classification and location standards for everyone, and per-industry sets such as clinical coding systems for healthcare, tariff codes and shipping terms for logistics, and technique and vulnerability catalogs for cyber insurance.
16 packs, and the domains they model
Each one is a deployable specification rather than a template folder. Where your domain is not listed, the layers above are the same, what changes is who authors the model.
Healthcare
Patients, encounters, claims, outcomes
Finance
Accounts, transactions, positions, exposure
Insurance
Policies, claims, reserves, agents
Cyber insurance
Threat actors, vulnerabilities, exposure, risk scoring
Manufacturing
Production lines, machines, materials, defects, work orders
Energy
Generation, consumption, grid, pricing
Oil & gas
Wells, reservoirs, production, safety
Utilities & water
Assets, service territory, outages, compliance
Retail
Products, stores, customers, transactions, promotions
Logistics
Shipments, routes, carriers, warehouses, containers
Maritime
Vessels, voyages, cargo, port calls
Aviation
Aircraft, maintenance, flights, telemetry
Telecom
Subscribers, usage, churn, network
Media
Content, channels, audience, campaigns, creators
Education
Students, courses, instructors, enrollment, assessment
Legal
Cases, filings, documents, matters, time entries
Agriculture
Crops, fields, seasons, equipment, harvest
Real estate
Properties, tenants, leases, valuation
The layer under the pack is the one that matters
Every vendor in this market ships accelerators. What a pack cannot give you on its own is a governance model, and that is the part underneath: the generated model still executes behind row and column policy, still emits lineage, still meters, and still inherits the classification ladder.
Which means adopting a pack does not create a second governance regime for the domain data. It is the same one.