AdResonance
Enter App
Developer Preview — 2026-04-22

Six layers around the operating record.

Publishing is one layer. Commerce, creative, CRM, analytics, and warehouses are five more. GENYS sits between all of them — not as a vendor aggregator, but as the operating record they feed into and execute through. AdResonance, the first vertical product built on GENYS, uses the Publishing-layer ad channels named below as its evidence examples; the six-layer taxonomy itself is generic.

This is a capability taxonomy. Not a vendor directory.

Vendors appear as evidence inside layers, not as the structure itself. Specific connector schemas, authentication models, and per-vendor limits are not part of the v1 public contract — they evolve inside each layer. The layers themselves are the stable claim.

Cross-layer flow

How a Decision moves through the layers.

Context flows in from Commerce, Creative, and CRM. A Decision is made, evaluated against Policy, and enqueued as a Workflow. Publishing executes out across ad channels. Analytics observes results back into the event-trace plane. Warehouses receive the longitudinal export. Every path threads through decision_id.

Context in

  • Commerce

    Inbound context. Product catalog, pricing, inventory, and conversion signal.

  • Creative

    Asset intake. Brand assets, design systems, component libraries, and template materials.

  • CRM

    Audience intake. Customer and lead identity, lifecycle state, lifetime-value signal.

Operating record

Decision · Policy · Workflow · Mutation Log

The contract surface. Context enters here as typed inputs; Policy evaluates; Workflow executes durably; Mutation Log records outcomes, all joined by one decision_id.

→ see /developers for shapes

→ see /architecture for runtime

Execution · feedback · export

  • Publishing

    Outbound execution. Where Decisions become channel-API calls.

  • Analytics

    Feedback observation. Performance, attribution, session quality, and outcome signal.

  • Warehouses

    Longitudinal export. Historical fact tables and analytical storage beyond the operational window.

Six layers

Each layer contributes something specific to the operating record.

A layer is not a vendor category — it is a capability role. Vendors inside a layer are evidence that the role is reachable; they are not the definition of what the layer is.

1

Publishing

Execution out

Outbound execution. Where Decisions become channel-API calls.

What it connects

Ad platforms that publish campaigns, bids, budgets, and creative to end audiences across search, social, video, shopping, and programmatic.

Why it matters

Every Mutation Log entry on /developers originates here. Channel-side changes are logged against the decision_id that caused them — the audit trail is load-bearing because Publishing is never the authority, only the execution surface.

Evidence

Google Ads, Meta Ads — examples, not the taxonomy.

2

Commerce

Context in

Inbound context. Product catalog, pricing, inventory, and conversion signal.

What it connects

Storefront, catalog, and supplier-network platforms that define what is being advertised — DTC storefronts, industrial catalogs, distributor feeds.

Why it matters

A Decision about a catalog-heavy vertical is only as accurate as the catalog it sees. Commerce ingestion is what makes long-tail SKU coverage, category-share allocation, and quote-driven routing reasonable rather than guessed.

Evidence

Shopify, EFP Catalog (industrial) — examples, not the taxonomy.

3

Analytics

Feedback loop

Feedback observation. Performance, attribution, session quality, and outcome signal.

What it connects

Analytics, attribution, and event-routing systems that close the loop from execution back to decision quality.

Why it matters

Outcomes return through this layer and feed the event-trace data plane. Without it, Memory has no ground truth and decision quality cannot be measured across time. This is the layer that makes 'decisions improve' a fact, not a claim.

Evidence

Google Analytics, Segment — examples, not the taxonomy.

4

Creative

Context in

Asset intake. Brand assets, design systems, component libraries, and template materials.

What it connects

Design tools and asset stores where creative production actually happens — design files, component systems, creative cloud libraries.

Why it matters

The refresh_creative Decision type and every generated variant draw from this layer. Approved-claim libraries and brand-governed assets are resolvable back to their source of truth, so governance survives creative production rather than being reviewed after it.

Evidence

Figma, Canva — examples, not the taxonomy.

5

CRM

Context in

Audience intake. Customer and lead identity, lifecycle state, lifetime-value signal.

What it connects

CRM, email-audience, and marketing-automation systems that carry identity and intent upstream of ads.

Why it matters

A rebalance_spend or pause_underperformer Decision is weighted by real qualified-buyer signal when CRM data is wired in — not just platform-side impressions. Quote-intent, deal stage, and RFQ activity land here, not in the ad channel.

Evidence

HubSpot, Salesforce — examples, not the taxonomy.

6

Warehouses

Longitudinal export

Longitudinal export. Historical fact tables and analytical storage beyond the operational window.

What it connects

Data warehouses and lakehouses that serve as the long-term analytical plane.

Why it matters

This is the destination of the Analytical data plane described on /architecture. Daily summaries, audit marts, and outcome rollups land here. It is what makes 'every override kept' meaningful over quarters, not just over weeks.

Evidence

Snowflake, BigQuery — examples, not the taxonomy.

Status by layer

Not every layer is equally live. This is the honest map.

Live means the out-of-the-box connector path ships end-to-end. Partial means some connectors are wired today, others are onboarded through engagement. Scaffolded means the layer exists in the runtime model but no generally-available connectors ship in v1 — onboarding happens through enterprise engagement.

LayerStatusNote
PublishingLiveSeven ad-platform OAuth flows wired today. This is the only layer where the out-of-the-box connector path ships end-to-end.
CommercePartialOne industrial-scoped commerce connector is live (catalog sync for manufacturing and parts environments). Broader e-commerce ingestion — Shopify, WooCommerce, BigCommerce, custom storefronts — is scaffolded and onboarded through engagement.
CreativePartialDesign-tool connectors for the two most-requested systems are live today. Cloud-library and file-store connectors are scaffolded.
AnalyticsScaffoldedThe feedback loop runs through platform-native analytics today. Third-party analytics and attribution connectors are scaffolded, not yet wired. This is a deliberate sequencing call — the event-trace plane comes first, the third-party loopbacks follow.
CRMScaffoldedThe largest v2 capability gap. No live CRM connectors. Audience ingestion happens via platform-side custom audiences today, not upstream from the CRM. Wiring this layer is how industrial and B2B demand signal becomes first-class.
WarehousesScaffoldedThe analytical-plane export path is scaffolded per /architecture. No live warehouse connectors in v1. Enterprise engagements can receive scheduled exports on request.

Internal connector schemas, authentication models, and per-vendor rate-limit behavior are deliberately not published in v1. A connector contract is committing work; we will publish it when it is stable enough to not break consumers. The /developers trust model covers the rest.

The platform is topology, not logos.

Vendors enter and leave. Layers stay. If your use case needs a connector not listed under Live today — industrial commerce, B2B CRM, warehouse exports, partner-specific analytics — the path in is enterprise intake, not a signup form.

Infrastructure for advertising capital.

Pay on what the system governs. Scale without upgrading.