The Impact of Generative and Agentic AI on Enterprise Application Architecture

Introduction

Enterprise applications are complex software systems that handle critical business processes within organizations. These applications often involve intricate data structures, persistent storage, high user concurrency, extensive user interfaces, and integration with other enterprise systems. The emergence of generative AI (GenAI) and agentic AI, including LLMs, is poised to significantly impact the design and implementation of enterprise applications.

The Evolving Role of Software Engineers

There is a lot of noise in the market about GenAI and agentic AI stealing jobs of software engineers. However, this is not necessarily the case. While these AI technologies can automate certain tasks, they also create new opportunities for software engineers to design and implement more intelligent, adaptive, and user-centric applications.

The role of software engineers is evolving from writing code to designing and managing AI-powered systems. Software engineers will need to develop new skills and knowledge to effectively leverage these AI technologies. They will need to understand how to design and implement AI agents, train and evaluate GenAI models, and integrate these technologies into enterprise applications.

This article explores how these AI technologies will influence various aspects of enterprise application architecture, including layering, domain logic, data source mapping, web presentation, concurrency, session state, and distribution strategies.

Layering

Traditional enterprise applications often employ a layered architecture, where each layer provides services to the layers above it and uses services from the layers below it. The three principal layers are the presentation layer, domain logic layer, and data source layer.

• Presentation Layer: This layer handles user interaction, displaying information, and interpreting user commands. GenAI can revolutionize this layer by enabling dynamic content generation, personalized user interfaces, and conversational interfaces powered by large language models (LLMs).
• Domain Logic Layer: This layer contains the business logic, including calculations, validations, and data manipulation. AI agents can automate complex decision-making processes, optimize workflows, and provide intelligent assistance to users within this layer.
• Data Source Layer: This layer communicates with external systems, such as databases, message queues, and other enterprise applications. GenAI can be used to generate synthetic data for testing and development, while AI agents can automate data integration and transformation tasks.

Organizing Domain Logic

Enterprise applications typically organize domain logic using patterns such as Transaction Script, Domain Model, or Table Module.

• Transaction Script: This pattern organizes business logic by procedures, where each procedure handles a single request from the presentation layer. AI agents can be used to orchestrate and execute these scripts, dynamically adapting to changing business rules and conditions.
• Domain Model: This pattern uses an object model of the domain that incorporates both data and behavior. AI agents can enhance domain models by providing intelligent recommendations, predictions, and anomaly detection capabilities.
• Table Module: This pattern uses one class to provide domain logic to a table or view in the database. AI agents can be used to optimize database queries, manage data access, and enforce data integrity rules within this pattern.

Mapping to Relational Databases

Object-relational mapping (ORM) is used to map objects in a domain model to tables in a relational database. Common ORM patterns include Table Data Gateway, Row Data Gateway, Active Record, and Data Mapper.

• Table Data Gateway: This pattern uses an object to act as a gateway to a database table. AI agents can be used to optimize data access through this gateway, dynamically adjusting query parameters and caching strategies based on usage patterns.
• Row Data Gateway: This pattern uses an object to act as a gateway to a single record in a data source. AI agents can be used to personalize data retrieval and presentation for individual users within this pattern.
• Active Record: This pattern wraps a row in a database table or view, encapsulates the database access, and adds domain logic on that data. AI agents can enhance Active Records by providing intelligent data validation, enrichment, and transformation capabilities.
• Data Mapper: This pattern uses a layer of mappers to move data between objects and a database while keeping them independent of each other and the mapper itself. AI agents can be used to optimize data mapping and transformation logic, dynamically adapting to changes in data schemas and business requirements.

Web Presentation

Web presentation patterns deal with how the application interacts with the user through a web interface. Common patterns include Model View Controller (MVC), Page Controller, Front Controller, Template View, Transform View, and Two Step View.

• Model View Controller (MVC): This pattern splits user interface interaction into three distinct roles: Model, View, and Controller. GenAI can enhance MVC by generating dynamic views based on user context and preferences, while AI agents can personalize the user experience by adapting the model and controller logic to individual users.
• Page Controller: This pattern uses an object to handle a request for a specific page or action on a website. AI agents can be used to optimize page navigation and content delivery based on user behavior and preferences.
• Front Controller: This pattern uses a controller to handle all requests for a website. AI agents can be used to provide centralized control over user interactions, enforcing security policies and adapting the application behavior to different user roles and contexts.
• Template View: This pattern renders information into HTML by embedding markers in an HTML page. GenAI can be used to generate dynamic templates based on user context and preferences, while AI agents can personalize the content within these templates.
• Transform View: This pattern processes domain data element by element and transforms it into HTML. AI agents can be used to optimize data transformation logic and adapt the presentation of data based on user needs and preferences.
• Two Step View: This pattern turns domain data into HTML in two steps: first by forming some kind of logical page, then rendering the logical page into HTML. AI agents can be used to optimize both steps of this process, dynamically adapting the logical page structure and HTML rendering based on user context and preferences.

Concurrency

Concurrency is the ability of an application to handle multiple simultaneous users accessing and modifying data. Concurrency control mechanisms are essential to ensure data integrity and prevent conflicts. Common concurrency patterns include Optimistic Offline Lock, Pessimistic Offline Lock, Coarse-Grained Lock, and Implicit Lock.

• Optimistic Offline Lock: This pattern prevents conflicts between concurrent business transactions by detecting a conflict and rolling back the transaction. AI agents can be used to predict and prevent conflicts before they occur, improving concurrency and reducing the need for rollbacks.
• Pessimistic Offline Lock: This pattern prevents conflicts between concurrent business transactions by allowing only one business transaction at a time to access data. AI agents can be used to optimize lock management, dynamically adjusting lock granularity and duration based on usage patterns and conflict probabilities.
• Coarse-Grained Lock: This pattern locks a set of related objects with a single lock. AI agents can be used to dynamically adjust lock granularity based on the context of user interactions, improving concurrency while ensuring data integrity.
• Implicit Lock: This pattern allows framework or layer supertype code to acquire offline locks. AI agents can be used to optimize lock acquisition and release strategies within this pattern, improving concurrency and reducing lock contention.

Session State

Session state refers to the data that is maintained for a specific user session. There are different ways to store session state, including Client Session State, Server Session State, and Database Session State.

• Client Session State: This pattern stores session state on the client, such as in cookies or hidden fields in a web page. AI agents can be used to personalize client-side session state, adapting the user interface and application behavior based on individual user preferences and context.
• Server Session State: This pattern keeps the session state on a server system in a serialized form. AI agents can be used to optimize server-side session state management, dynamically adjusting session timeout values and caching strategies based on usage patterns and resource availability.
• Database Session State: This pattern stores session data as committed data in the database. AI agents can be used to manage database session state, ensuring data consistency and integrity across multiple user sessions.

Distribution Strategies

Distribution strategies deal with how the application is deployed across multiple servers or nodes. Common distribution patterns include Remote Facade and Data Transfer Object (DTO).

• Remote Facade: This pattern provides a coarse-grained facade on fine-grained objects to improve efficiency over a network. AI agents can be used to optimize remote communication, dynamically adjusting communication protocols and data serialization strategies based on network conditions and user context.
• Data Transfer Object (DTO): This pattern uses an object to carry data between processes in order to reduce the number of method calls. AI agents can be used to optimize DTO design and usage, dynamically adapting the structure and content of DTOs based on user needs and context.

Conclusion

• The Transformative Impact of Agentic AI on Enterprise Applications

Agentic AI, with its ability to learn, adapt, and make decisions, is poised to revolutionize enterprise applications. By embedding intelligence within applications, agentic AI can streamline processes, enhance user experiences, and unlock new levels of efficiency and innovation.

• LLMs for Intuitive User Interfaces and Automation

Large Language Models (LLMs), with their advanced natural language processing capabilities, can transform user interfaces. LLMs enable conversational interfaces, allowing users to interact with applications using natural language, making complex systems more accessible and user-friendly. Furthermore, LLMs can automate tasks, such as data analysis, report generation, and decision-making, freeing human workers to focus on more strategic initiatives.

• GenAI for Content Creation and Innovation

Generative AI (GenAI) empowers enterprise applications with the ability to create new content, designs, and solutions. GenAI can generate marketing materials, product designs, and even software code, accelerating innovation and development processes. Moreover, GenAI can personalize user experiences by tailoring content and recommendations to individual needs and preferences.

• AI Agents for Intelligent Automation and Optimization

AI agents, with their ability to perceive, reason, and act autonomously, can automate complex tasks and optimize workflows within enterprise applications. AI agents can monitor system performance, identify anomalies, and proactively take corrective actions, ensuring smooth and efficient operation. Additionally, AI agents can provide intelligent assistance to users, guiding them through complex processes and offering personalized recommendations.

• A Holistic Transformation of Enterprise Application Architecture

The combined capabilities of LLMs, GenAI, and AI agents will impact all aspects of enterprise application architecture. From layering and domain logic to data source mapping, web presentation, concurrency, session state, and distribution strategies, these AI technologies will drive a fundamental shift towards more intelligent, adaptive, and user-centric applications.

By embracing these technologies and adopting new design patterns, organizations can unlock new levels of efficiency, innovation, and user satisfaction in their enterprise applications. The future of enterprise applications lies in harnessing the power of agentic AI to create intelligent systems that empower users, streamline processes, and drive business growth.

Disclaimer: This publication contains general information and is not intended to be comprehensive nor to provide professional advice or services. This publication is not a substitute for such professional advice or services, and it should not be acted on or relied upon or used as a basis for any investment or other decision or action that may affect you or your business. Before taking any such decision you should consult a suitably qualified professional advisor. While reasonable effort has been made to ensure the accuracy of the information contained in this publication, this cannot be guaranteed, and neither associated organization nor any affiliate thereof or other related entity shall have any liability to any person or entity which relies on the information contained in this publication. Any such reliance is solely at the user’s risk. This article may contain references to other information sources. Views are personal.