AI-Augmented Solution Architecture (ASA+)

Overview

ASA+ (AI-Augmented Solution Architecture) adopts a more augmented, integrative, absorptive, evolutionary, hybrid, or fused approach to AI enterprise solutions, compared with the AI-first orientation of ASA approach (AI-Native Solution Architecture, see this link). It is positioned as an enterprise AI absorption and coexistence architecture, aiming to maintain architectural continuity while enabling AI augmentation.

ASA+ Architectural Approach

ASA+’s architectural approaches include:

ASA+ Modeling Elements

As an AI-augmented architectural approach, ASA+ incorporates both non-AI elements and AI-specific elements (which are heavily emphasized in AI-native architecture). As a result, ASA+ operates on a mixed set of AI and non-AI elements, including a shared subset with the base ASA approach. Table 1 presents the primary AI and non-AI elements of ASA+.

Element Name AI-Specific Definition
Access Interface   Represents the interaction channels, UI/UX surfaces, and entry points through which humans engage with the solution.
Application   Represents a bounded software system, enterprise application, or business component that integrates with or consumes AI capabilities.
App Logic   Represents explicitly defined non-GUI logic, control flow, or compositional behavior of an application.
Data Service   Represents services responsible for data access, integration, transformation, federation, and transactional integrity.
Technical Component   Represents reusable technical capabilities, utility services, and cross-cutting infrastructure functions available across the solution.
AI Agent Yes Represents an autonomous AI entity capable of goal-directed reasoning, planning, and action.
AI Coordinator Yes Represents the coordination logic, workflow control, and multi-agent management that sequences and routes AI operations.
Context State Yes Represents the mechanisms for managing conversational state, memory, prompt engineering, and interaction coherence.
AI Model Yes Represents the models, inference engines, and reasoning frameworks that generate predictions, decisions, or outputs.
Knowledge Service Yes Represents the semantic retrieval, RAG, embedding, and knowledge management capabilities that ground AI responses in relevant information.
AI/ML Lifecycle Yes Represents the lifecycle management processes for model training, experimentation, versioning, and deployment.
Autonomous Tool Yes Represents external functions, plugins, and third-party services that extend AI capabilities through invocation.

Table 1: Primary ASA+ Elements

For the full list of ASA+ modeling elements and its foundational specification, refer to this link.

ASA+ architectural services can be categorized into three types: fully autonomous applications, agentic applications with varying degrees of autonomy, and deterministic automation applications.

ASA+ Example

Here are examples of ASA+ modeling cases.

Canonical Case Example

Figure 1 illustrates an ASA+ for an AI-Augmented Enterprise Operations Platform. This example demonstrates:

Canonical Case Example

Figure 1: ASA+ for an AI-Augmented Enterprise Operations Platform

Edge Case Example

Figure 2 shows ASA+ edge case for a High-Risk Human-Governed AI Decision Environment. This edge case demonstrates:

Edge Case Example

Figure 2: ASA+ Edge Case Example

ASA+ uses its model specification and maintains a close relationship with AI-native solution architecture.

For the relationship and relevance among ASA model and approach, and ASA+ approach, see this link.