The Impact of Generative AI on Enterprise Architecture

Enterprise architecture is a broad field and organizations hire skilled contributors to fulfill a subset of its job function. Some hire for managing business architecture while others hire for managing their application and tech architecture. When an Enterprise Architect works for sometime under these employers, they seem to lose sight on the actual job duties of an Enterprise architect. This leads to unfair performance ratings, unbalanced contribution to business and technology needs. For someone who I thought I was, even I fell a prey to one of my employers. I want to use this opportunity to help build the basics and proceed with actual subject of this blog.

I lost my job in the recent past because I saw the signs way too late. After a grueling grieving processes and chronic depression, I finally took complete accountability for the fall out and promised myself to never be in the same spot ever again.

I want to spend my time and resources to help my fellow architects in this community on the true job duties of an EA. I hope this blog provides the needed insights to my fellow Enterprise Architects and leaders alike because knowing what to expect from an Enterprise Architect is equally important than knowing one’s job duties.

Duties of an Enterprise Architect

During my grieving phase of my job loss, I read many books, blogs, spoke to countless experts in the field. I have summarized my findings below:

1. Aligning IT with Business Strategy: An EA is responsible for ensuring that the IT strategy and architecture align with the business’s overall strategy. This involves understanding both the business’s current and future needs irrespective of whether the business leaders have curtesy to invite you to key stakeholder meetings like strategy meetings.
2. Stakeholder Communication: Effective communication with stakeholders, including business leaders, is crucial. An EA must articulate the value of IT in achieving business goals and be involved in strategic conversations. Given that IT hires Enterprise Architects, I believe what this point highlights is the requirement to know the stakeholder list on any business capabilities.
3. Developing and Maintaining Architectural Frameworks: EAs are tasked with developing and maintaining frameworks that support efficient and effective IT system deployment, which aligns with the business model. Even though there are many popular frameworks out there like TOGAF, Zachman etc. it is important to customize the framework to your organizational needs. An EA who is skilled in this will be valued.
4. Guiding Technology Decisions: EAs guide the selection and implementation of technologies based on their alignment with business objectives and architectural principles. More specifically, an EA who is skilled in visualizing their business as an interaction between multiple domains and for each domain, has an ability to define a Destination architecture will be indispensable in the marketplace.
5. Ensuring Compliance and Risk Management: They ensure that IT systems comply with legal and regulatory requirements and assess and mitigate risks associated with IT architectures. There will be dedicated teams like Legal, PCI, auditing group etc. An EA must work closesly with these key members and incorporate their best practices into the proposed Destination architecture.
6. Driving Innovation: An EA should be at the forefront of exploring new technologies and approaches that can drive business growth and efficiency. An EA must never feel shy to dive head-first in upcoming technologies. Keeping oneself sharp in the field of technology will provide an EA with unprecendted strategic advantage. (See, I spent far too much time here while shifting my weights almost completly to this. Do be me).
7. Change Management: They play a key role in managing change, ensuring that technological changes align with business strategies and are adopted smoothly across the organization. Note that Change management is the most complex task in an EAs job duties because if there is one thing that unites all of the humans under one umbrella it is the fact that we all hate changes. Make sure the proposed change has a valid rationale and that it aligns with the strategic imperative of the business. Make sure the change is linear and easy to adopt. Break the process to granular steps wherever appropriate to improve the proposed adoption.

Now that I have shared the job duties of an EA, I should confess to you all that I am a fan of TOGAF framework. My love for TOGAF framework is its modularity. Any EA can literally read the framework, its process and create their own process without having to break the existing process. The simplicity in the adoption far outweighs the complexity in adopting any other frameworks. Therefore, I have structured the rest of this blog around the Architecture Development Method as proposed by the TOGAF framework.

I asked ChatGPT to provide me a TL;DR (Too Long, Didn’t Read) version of TOGAF framework and it did a decent job summarizing it. I am going to share it AS-IS so that I can take you on to the actual topic of this blog.

TOGAF (The Open Group Architecture Framework)

It is an enterprise architecture framework that provides a comprehensive approach to designing, planning, implementing, and governing an enterprise information architecture. It is widely used for optimizing business efficiency in a systematic and disciplined way. Key components include:

1. ADM (Architecture Development Method): The core process of TOGAF, providing a step-by-step approach to developing and managing enterprise architecture.
2. Content Framework: Describes the types of architectural artifacts (like models and diagrams) that should be produced during ADM phases.
3. Enterprise Continuum: A virtual repository of all the architectural assets (models, patterns, architecture descriptions) that exist both within the organization and in the wider architecture community.
4. Reference Models: Includes TOGAF Foundation Architecture and the Integrated Information Infrastructure Reference Model (III-RM), providing a set of common architectural elements.
5. Architecture Governance: A practice for ensuring the compliance of the enterprise architecture with organizational standards and objectives.

TOGAF is notable for its emphasis on aligning IT with business goals, adaptability to a wide range of organizational contexts, and its iterative process that allows for continuous improvement of the architecture.

The Architecture Development Method (ADM)

The Architecture Development Method (ADM) is a central aspect of The Open Group Architecture Framework (TOGAF). It serves as a comprehensive approach to designing, planning, implementing, and governing enterprise information architecture.

Why ADM is Needed

• Structured Approach: ADM provides a systematic, step-by-step process to develop and manage enterprise architectures. It ensures that all aspects of an organization’s architectural needs are addressed in a methodical way
• Alignment with Business Goals: By starting with a clear understanding of the organization’s vision, strategies, and business goals, ADM ensures that the architecture is closely aligned with what the business needs
• Stakeholder Collaboration: The method promotes collaboration among various stakeholders, including business owners, developers, and IT professionals, ensuring that all perspectives are considered in the architecture.
• Risk Management: ADM helps in identifying and managing risks at every stage of the architecture development process.

Now that you understand the basics of TOGAF and its underlying ADM, let us envision a virtual organization and its core business model. For this organization, let us see how Generative AI can help an Enterprise Architect to generate the needed artifacts for each phase of the ADM.

You can use similar approach to any of the frameworks out there and it should still work.

Our Imaginary Business

Our imaginary business is a new finance startup, that creates a value stream to its customers through their credit card offerings. The startup has chosen TOGAF as their initial framework of choice and has tasked you the EA, to generate various artifacts required to build their business model and technology model.

The impact of Generative AI in each phase of Architecture Development Method

As an Enterprise Architect, your job is to run the design and run the ADM as needed for your iteration. Let us see each phases in detail.

1. Preliminary Phase

The Preliminary Phase in TOGAF’s ADM primarily serves to prepare the organization for successful architecture development. It involves defining the scope, setting up the architecture.

Use Case:

Setting up an enterprise architecture team for a credit card issuing finance organization.

Artifact:

Enterprise Architecture Framework document.

The EAF is a document that contains every single artifact and asset, models and patterns that you will develop for your organization. Usually, there are tools to manage this in its entirety or teams choose Content Management Systems like Confluence to start storing the outputs.

Significance:

Establishes the tailored TOGAF framework, principles, and tools to be used throughout the architecture process.

2. Phase A: Architecture Vision

Phase A, Architecture Vision, in TOGAF’s ADM is focused on defining the high-level vision of the end architecture outcome. It aims to align this vision with the business goals and strategy, ensuring stakeholder buy-in and establishing the scope for the architecture work.

Use Case:

Envisioning a new cloud-based credit card processing system using AWS services.

Artifact:

Statement of Architecture Work.

Significance:

Outlines the scope, stakeholders, business goals, and constraints of the cloud-based system.

3. Phase B: Business Architecture

Phase B, Business Architecture, in TOGAF’s ADM, is dedicated to developing an in-depth understanding of the business strategy, governance, organization, and key business processes. Its purpose is to ensure that the business architecture is aligned with the organization’s objectives and effectively supports its operations and development.

Use Case:

Analyzing existing business processes for credit card application and approval.

Artifact:

Business Process Diagram.

A Business Process Diagram is a visual representation that outlines the sequence of activities, decisions, and flows within a business process. It serves to illustrate how various tasks are interconnected and coordinated to achieve specific business objectives, often including roles, responsibilities, and key decision points.

Significance:

Illustrates the current business processes and identifies areas for improvement.

4. Phase C: Information Systems Architectures

Phase C, Information Systems Architectures, in TOGAF’s ADM, involves developing architectures for an organization’s data and applications. This phase focuses on creating blueprints that detail how data is processed and managed, and how applications are designed and interact, to support the business architecture and requirements.

Data Architecture

Phase C, Data Architecture, in TOGAF’s ADM, specifically focuses on defining the structure, storage, management, and flow of data within an organization. This phase aims to align the data architecture with business requirements, ensuring effective data management and utilization for decision-making and operations.

Use Case:

Designing a data model for customer and transaction data.

Artifact:

Conceptual Data Model.

Significance:

Provides a high-level view of data entities and relationships relevant to the credit card business.

Applications Architecture

Phase C, Application Architecture, in TOGAF’s ADM, is dedicated to outlining a blueprint for the individual software applications and their interactions within the enterprise. Its aim is to ensure these applications support business processes effectively and are aligned with the organization’s technology strategy and objectives.

Use Case:

Planning the application landscape for managing credit card operations.

Artifact:

Application Interaction Diagram or Domain Interaction diagram.

Significance:

Shows how different applications interact and support business processes.

5. Phase D: Technology Architecture

Phase D, Technology Architecture, in TOGAF’s ADM, focuses on defining the hardware, software, and network infrastructure needed to support the deployment of core business applications. This phase aims to design a robust and scalable technology landscape that aligns with the organization’s data, application, and business architectures.

Use Case:

Selecting AWS services for hosting the credit card processing system.

Artifact:

Technology Stack Diagram.

Significance:

Identifies the AWS services and other technologies required to support the IS architectures.

6. Phase E: Opportunities and Solutions

Phase E, Opportunities and Solutions, in TOGAF’s ADM, is about identifying and evaluating potential delivery vehicles (projects, initiatives, or programs) that can realize the architecture vision. It aims to create a strategy for leveraging opportunities and addressing gaps between the current and desired states of the architecture.

Use Case:

Identifying opportunities to integrate new fintech solutions.

Artifact:

Opportunities Catalog.

An Opportunities Catalog in enterprise architecture, particularly within TOGAF, is a documented list of potential projects, initiatives, or changes that can help achieve the organization’s architectural vision and objectives. It typically includes information on the scope, impact, benefits, and priorities of each opportunity, aiding in strategic decision-making and roadmap development.

Significance:

Lists potential projects or solutions that could realize the architecture vision.

7. Phase F: Migration Planning

Phase F, Migration Planning, in TOGAF’s ADM, involves creating a detailed plan to move from the current architecture to the desired future state. This phase focuses on scheduling, resourcing, and coordinating the various activities and projects needed to implement the new architecture efficiently and effectively.

Use Case:

Planning the transition from the existing system to the new AWS-based system.

Artifact:

Implementation and Migration Plan.

Significance:

Provides a detailed roadmap for implementing the new system, including timelines and resources.

8. Phase G: Implementation Governance

Phase G, Implementation Governance, in TOGAF’s ADM, is focused on overseeing and guiding the actual implementation of the architecture. Its primary purpose is to ensure that the deployment adheres to the defined architectural plans, standards, and principles, and achieves the intended business objectives.

Use Case:

Overseeing the implementation of a secure payment gateway.

Artifact:

Governance Framework.

A Governance Framework document typically contains guidelines and procedures for decision-making, compliance monitoring, and the management of enterprise architecture initiatives. It outlines roles, responsibilities, processes, and metrics for effectively governing the implementation and evolution of the architecture.

Significance:

Ensures the project aligns with the architecture vision and compliance standards.

9. Phase H: Architecture Change Management

Phase H, Architecture Change Management, in TOGAF’s ADM, is dedicated to the ongoing management and monitoring of changes to the architecture. Its purpose is to ensure that the architecture remains relevant and effective in meeting business needs, even as those needs evolve over time.

Use Case:

Managing changes during the system’s lifecycle.

Artifact:

Change Request Log.

Significance:

Records and tracks changes to the architecture to ensure controlled evolution.

10. Requirements Management

The purpose of Requirements Management in TOGAF’s ADM is to manage and maintain a repository of all the requirements that the architecture must satisfy. This ongoing process ensures that these requirements are identified, prioritized, and addressed throughout the architecture development lifecycle.

Use Case:

Continuously capturing and managing stakeholder requirements.

Artifact:

Requirements Impact Analysis.

Significance:

Assesses the impact of changing requirements on the architecture.

Sample Prompts that can be given to Image Generation AI applications to Generate the needed artifacts

For the steps stated earlier, I created few prompts that could be used to generate a sample workflow. This can be used as a motivation to build the actual visualization.

Note: DALL.E, in my opinion, is an eye candy image generator. It should NEVER be used for technology visualizations its current state. Use these prompts on products that are made for these purposes and you will see better results.

Business Process Diagram for Credit Card Application

• Prompt: “Create a detailed business process diagram showing the steps involved in the credit card application and approval process within a financial institution. Include decision points, roles involved, and typical timeframes.”

Conceptual Data Model for Customer and Transaction Data

• Prompt: “Generate a conceptual data model that includes entities such as Customer, Credit Card, Transaction, and Payment. Show relationships and key attributes for each entity relevant to a credit card system.”

3. Application Interaction Diagram for Credit Card Operations

• Prompt: “Illustrate an application interaction diagram for a credit card management system, showcasing the interaction between CRM, transaction processing, risk assessment, and customer support applications.”

Here again, Dall-E created a fun looking image. The professional image will be different and should be drawn with professional tools like Visio etc.

4. Technology Stack Diagram Using AWS Services

• Prompt: “Design a technology stack diagram for a cloud-based credit card processing system using AWS services. Include components for database management, application hosting, security, and analytics.”

5. Governance Framework for Payment Gateway Implementation

• Prompt: “Develop a governance framework outline for overseeing the implementation of a secure payment gateway in a financial institution. Highlight key governance areas such as compliance, security standards, and stakeholder engagement.”

Tools that can be used to Generative artifacts with help from AI

I love working with the following tools during my EA ideation phase. These are powerful and an organization that invests in a robust EA, will have one or more of these tools at your disposal.

• Lucidchart — For creating process diagrams, data models, and architecture diagrams.
• Miro — Useful for brainstorming, creating architecture vision boards, and planning.
• Tableau or PowerBI — For data visualization and analysis.
• Microsoft Visio — For detailed business and IT architecture diagrams.

Open Source Tools

There are a few notable tools and libraries that can be leveraged for certain aspects of generative AI:

GPT-3 Models:

While OpenAI’s GPT-3 itself is not open-source, there are open-source alternatives like GPT-Neo and GPT-J, which are developed by EleutherAI. These models can be used for generating text-based content.

DALL-E:

OpenAI’s DALL-E, an AI model known for generating images from textual descriptions, is not open-source. However, there are initiatives like OpenAI’s DALL·E mini (now Craiyon) that offer similar functionalities with certain limitations.

TensorFlow and PyTorch:

These are open-source machine learning libraries that can be used to build custom generative models. They require significant expertise in machine learning and model training.

Diagrams.net (formerly draw.io):

While not specifically a generative AI tool, diagrams.net is an open-source diagram software that can be used to manually create various types of diagrams and architectural artifacts.

Apache Superset:

An open-source data visualization tool that can be used to create dashboards and visualizations, useful for presenting data architecture and analytics.

Pencil Project:

An open-source GUI prototyping tool that can be useful for sketching out application interfaces or workflows.

MLflow:

An open-source platform for managing the machine learning lifecycle, including experimentation, reproducibility, and deployment of ML models.

Conclusion

Enterprise Architecture is a broad field with broader responsibilities, Generative AI tools are increasingly redefining the way we generate artifacts even amidst pressure and lack of decent data. Image generation tools in Generative AI are at infancy. Therefore, not all tools will generate an Enterprise worthy artifact. Care should be taken to use AI as a co-pilot rather than equipping it with roles of autonomy.

By carefully choosing these tools, an Enterprise Architect can successfully map their business and can use that to create strategic advantage to their leaders.

Generative AI will definitely be a co-pilot in creating artifacts as needed by an EA. Knowing when to interfere manually with the generated content is a key skill that Enterprise Architects will have to develop in the near future.