Unlocking Effectiveness: How Generative AI is Revolutionising Enterprise Architecture

Generative AI limitations

Generative AI is a type of AI capable of creating new and novel outputs, such as text, code, images, videos and audio. But generative AI is unaware of truthfulness, as it is optimised to predict the most likely response based on the context of the current conversation, the prompt provided and the data set it is trained on.

Since the large language models (LLMs) underpinning generative AI are pre-trained, and it is (for now) cost-prohibitive to retrain the models with every change in EA data, this means that the context and prompt must ensure that data returned from the model reflects known EA knowledge and does not make up knowledge it doesn’t have.

Since LLMs have limited size in terms of the context they can deal with, and consequently cannot include the full corpus of EA knowledge as context, this means LLMs need to be used in very targeted, intentional ways for EA purposes to ensure reliable, consistent outputs.

The role of enterprise architecture

Enterprise architecture’s primary purpose is to enable alignment of an organisation’s business strategy and operational needs with its technology-enabled capabilities and infrastructure. Misalignment between Information Technology (IT) and business needs hampers the ability to respond swiftly, affordably and securely to critical business demands.

To help achieve this alignment, individuals with the role of enterprise architect enable execution and delivery of the organisation’s strategy and vision through either existing or innovative technology capabilities. They do this by cataloging and mapping aspects of an enterprise that are often not well understood – for example, by mapping the organisation’s strategy and structure to the catalog of processes, information and technical assets.

Those in enterprise architecture roles therefore have a responsibility to steward the overall technology strategy, and as such must understand the technological readiness of an organisation wishing to adopt and embed GenAI into its overall architecture. An enterprise architecture function can leverage the collective input from architects across the organisation to provide AI leaders with readiness viewpoints on:

• Third-party AI technology
• Build/borrow/buy AI technology selection/decisions
• Data quality
• Infrastructure services
• Integration architecture

Mature EA practices with a strong business architecture capability can drive executive discussions around the strategic and social implications of using GenAI for critical business operations by providing organisational viewpoints and perspectives. Its activities may highlight previously unarticulated dependencies and potential unintended consequences. Such insights can also help with prioritisation and sequencing decisions for AI investments, which are highly dependent upon existing data and analytics maturity.

Those in enterprise architect roles can also provide insights that help AI leaders assess the technology strategy impact of injecting GenAI into key enterprise processes. Enterprise architects can work with AI leaders to decide:

• Which business capabilities GenAI would offer competitive advantages for,
• Which capabilities will be table stakes with respect to the wider industry, and
• Which capabilities will be commoditised or non-differentiating and best delivered through managed services or third-party providers.
• These decisions constitute the technology strategy as it relates to AI, acting as a north star for technology decision-making.

The impact of GenAI on EA activities

GenAI affects the enterprise architecture in multiple ways, many which cannot yet be predicted. For the purposes of this article, the focus is on an enterprise architect’s role in:

• Facilitating strategic technology AI investment decision-making.
• Ensuring sustainable alignment between business objectives and IT capabilities.
• Governing and managing the technology underpinning GenAI initiatives.

Facilitating strategic technology AI investment decision-making

The focus here is less on technology solutions and more on business processes and capabilities and the potential for transformative improvements with the introduction of GenAI.

This work involves typical business architecture activities such as cataloging end-to-end processes aligned with customer experience, identifying objectives and key results with respect to introducing AI capabilities (such as customer retention, revenue improvements, productivity improvements, compliance improvements, etc.) and aligning those objectives with strategic decisions to differentiating investments, table-stake investments and non-differentiating activities that could be outsourced.

To enable rapid collation and alignment of information from disparate sources, business architects leverage industry-specific reference architectures and use GenAI to accelerate the process of mapping associations across objectives, processes and capabilities, enabling the creation of a meaningful enterprise view that can be used as input into technology strategy decision-making activities, including potentially non-obvious dependencies to identify areas of maximum investment opportunity.

Ensuring sustainable alignment between business goals and IT capabilities

A key benefit of GenAI for EA will be to reduce the toil associated with activities that those in enterprise architecture roles need to do but often do not have the time or resources to do effectively. This includes:

• Gathering and cleaning necessary data from IT sources
• Identifying and engaging with key stakeholders
• Identifying and communicating architectural insights and options
• Performing architectural reviews
• Identifying relevant existing technology capabilities across the enterprise
• Providing and maintaining architectural diagrams, viewpoints and perspectives and more.

Consequently, GenAI for EA will help significantly accelerate the democratisation of enterprise architecture activities and responsibilities to the parts of the organisation that are closest to the needs.

For example, GenAI opens the door to future automation of some aspects of architecture reviews, allowing stakeholders to have an AI-based process assess proposed architectures and address any obvious gaps to making recommendations as to whether a formal human review is needed.

Once the technology strategy with respect to AI has been established, enterprise architects work with the project management office to scope the most appropriate technology projects to deliver the AI strategy, so that well-formed IT projects can be scoped, funded and prioritised.

Part of this process is to understand the ‘architecture runway’ necessary to ensure the technological readiness of systems to deliver AI objectives, as described above. GenAI can accelerate the process of understanding the AI readiness of applications, data and infrastructure through accelerated mapping of business architecture information to these architectural dimensions so that the architect can communicate technical dependencies and gaps to stakeholders.

AI model lifecycle governance

GenAI can also help enterprise architects with their role in model lifecycle management – the process of planning, implementing, maintaining and evolving LLMs within an organisation’s IT infrastructure and overarching AI strategy. Knowing which applications use (or plan to use) which versions of which models is key to enabling effective AI governance, and EA can play a key role here in making useful insights available to AI leaders.

Generative AI challenges and risks

Given the substantial socio-technical risks associated with the use of GenAI for enterprise purposes, AI governance is a critical capability that organisations must develop. Protiviti’s AI governance framework identifies best practices relating to AI governance principles, use case development and business value.

Enterprise architecture, as the primary vehicle through which technology strategy execution is governed by IT, has a role in contributing data and insights to the application of AI governance best practices, as well as ensuring that AI governance objectives are incorporated into architectural principles and standards assessed during software planning and delivery.

EA automation outlook

With enterprise IT complexity expected to continue to rise exponentially, organisations’ expectations of the value of enterprise architecture cannot, practically speaking, be met in the future without automation. Increasingly, that automation is taking the form of GenAI working in collaboration with “knowledge graphs.” A knowledge graph is a digital map that shows how different information concepts are related to each other and is humanly curated rather than derived by statistical inference. As such, knowledge graphs can represent facts, not probabilities.

Traditional architecture modeling tools, crucial for designing complex systems, will have features augmented by AI and knowledge graphs. Enterprise architecture management tools will benefit from higher quality data from AI-enhanced IT processes, reducing the need for AI to address IT process data quality issues at the point of EA consumption, while significantly improving the utility of data in less structured EA activities, such as business/technology planning.

What’s ahead

Generative AI is set to revolutionise, not render obsolete, the field of enterprise architecture by empowering those in enterprise architecture roles to deliver more value. Enterprise architects are recognised stewards of technology strategy, regardless of where they are in the organisation and what other roles they may have. By democratising informed technology decision making, resulting in significantly reduced central decision-making and IT governance lead-times, strategic business and technology AI objectives can be achieved more swiftly, affordably, responsibly and intentionally.

To learn more about our artificial intelligence services, contact us.