What Is the Four-Layer RaaS Stack?

The Four-Layer RaaS Stack is the Crown Point Advisory Group architectural framework for the infrastructure required to execute Resolution as a Service (RaaS) at enterprise scale. It defines four sequential layers, each a prerequisite for the one above it, that together constitute the Outcome-First backbone of a RaaS-native platform. A vendor who has built all four layers can prove every resolution claim. A vendor who has built only the top layers without the foundation cannot.

Layer 1: Physical Edge Runtime

The foundation of the stack is the Physical Edge Runtime: the compute infrastructure, model hosting, and orchestration layer where AI agents actually execute. In the RaaS architecture, this layer must be instrumented from the ground up to capture cost-to-serve at the resolution level.

A traditional SaaS infrastructure layer tracks uptime, latency, and error rates. A RaaS infrastructure layer must also track tokens consumed per resolution attempt, orchestration cost per agent action, model inference cost by resolution type, and the compute signature that distinguishes autonomous execution from human-assisted execution.

This instrumentation is not optional. Without it, the vendor cannot apply the 1-to-4 Rule, cannot detect power user margin erosion, and cannot produce the cost-to-serve data that Phase 2 of the Three-Phase RaaS Transition Roadmap requires. Vendors who attempt to build RaaS pricing on top of an uninstrumented infrastructure layer are pricing from assumption rather than evidence.

Layer 2: High-Fidelity Repository

The second layer is the High-Fidelity Repository: the graph-structured institutional knowledge architecture that gives AI agents the domain context, audit trails, and attribution records needed to execute resolutions with accuracy and auditability.

The Repository is the most strategically significant layer of the stack because it is the primary competitive moat. The Physical Edge Runtime can be replicated by any vendor with sufficient infrastructure investment. The High-Fidelity Repository cannot be replicated without access to the same institutional data over the same period of time.

Layer 2 is also where attribution, the second of the three Atomic Resolution criteria, is technically enforced. Every resolution execution generates an audit record in the Repository: what the agent accessed, what it decided, what action it took, and what outcome resulted. That audit record is the evidentiary foundation of every invoice.

Layer 3: Agentic Connectivity

The third layer is Agentic Connectivity: the integration infrastructure that allows AI agents to reach across the enterprise, accessing data and triggering actions in the customer’s broader system environment.

The primary technical standard for this layer is Model Context Protocol (MCP), the open standard developed by Anthropic that allows agents to connect directly to the APIs and data layers of external platforms. Layer 3 is where SaaS Demotion either becomes an existential threat or a competitive advantage, depending on whether the vendor has built the Repository and Runtime layers below it.

A vendor with a mature Layer 2 Repository and an MCP-connected Layer 3 is an authoritative resolution engine: agents that connect to its data layer have access to the richest domain context available, which produces higher quality resolutions, lower reopen rates, and stronger attribution records than any competitor with a shallower Repository.

A vendor without a mature Layer 2 who nonetheless opens Layer 3 MCP connectivity is accelerating SaaS Demotion by giving agents direct data access without the compensating revenue model or the attributable resolution quality to defend the invoice.

Layer 3 also defines the agent permission architecture: read-only agents that observe and report, recommendation agents that suggest but require human approval, execution agents that act autonomously within defined resolution types, and escalation triggers that route judgment-sensitive decisions to human operators. This permission hierarchy is the technical implementation of RaaS Stewardship.

Layer 4: Outcome Interface

The fourth layer is the Outcome Interface: the commercial layer that translates resolution execution into verifiable billing claims. It includes the resolution counter, the quality gate (the 48-hour reopen monitor that determines whether a completed resolution holds), the invoice generator, and the audit trail that customers can inspect to verify what they are being charged for.

Layer 4 is where the measurement trust infrastructure lives. A customer who disputes an invoice needs to see, from the audit trail, exactly which resolutions were completed, when they were completed, what the agent did, and why the vendor claims they were autonomous rather than human-assisted. A vendor who cannot produce that record cannot defend a RaaS invoice.

The Outcome Interface is also where the Resolution Dispute Protocol operates. Type 1 disputes, failed resolutions reopened within 48 hours, are detected automatically at Layer 4 by the quality gate. Type 2 disputes, misclassified resolutions where a human-executed resolution was billed at the AI rate, are resolved by cross-referencing the Layer 4 invoice against the Layer 2 attribution records.

The Dependency Structure

The four layers have a strict dependency structure. Layer 2 cannot operate without the instrumented Runtime of Layer 1. Layer 3 cannot produce attribution-quality resolutions without the Repository of Layer 2. Layer 4 cannot generate defensible invoices without the audit records from all three layers below it.

This dependency structure is why the Phase 2 Hybrid Pilot in the Three-Phase RaaS Transition Roadmap requires all four layers to be instrumented before any customer is moved to outcome pricing. Vendors who attempt to price outcomes before the full stack is instrumented are building the invoice on top of an evidentiary gap. The customer’s procurement team will find that gap. The reopen rate will expose it if the procurement team does not.

The Stack and the 2024 Workflow-First to 2026 Outcome-First Transition

The Four-Layer RaaS Stack is the architectural answer to the transition described in the RaaS Manifesto from the 2024 Workflow-First model to the 2026 Outcome-First model.

Under the Workflow-First model, the primary interface was the proprietary vendor UI, data was stored in relational siloed structures, logic was executed by humans through the interface, and the system moat was UI friction and vendor lock-in. Under the Outcome-First model, the primary interface is the autonomous agent layer, data is graph-based in the High-Fidelity Repository, logic is executed agentically, and the system moat is connectivity, context, and data integrity.

The Four-Layer Stack is the technical specification for building the Outcome-First model from the ground up, or for rebuilding toward it from a Workflow-First starting point.

The Four-Layer RaaS Stack is defined in Chapter 6 of the Crown Point Advisory Group RaaS Manifesto. The operational build sequence for each layer is in the Vendor Transition Playbook, organized by phase and workstream. The High-Fidelity Repository build roadmap is detailed separately at What Is the High-Fidelity Repository?