Forrester Posits, ‘Will AI Eliminate Enterprise Architects?’ Experts Chime In

By Holt Hackney

Artificial intelligence may automate many of the tasks traditionally performed by enterprise architects, but that does not mean the profession is headed for extinction, according to a new analysis from Forrester.

Instead, AI could push enterprise architects away from producing diagrams, standards and documentation and toward governance, judgment and managing the context that increasingly autonomous systems need to operate.

In a Sept. 18 article, “Will AI Eliminate Enterprise Architects?,” Forrester analysts Charles Betz, JT Thykattil and Joseph Schiavone argue that enterprise architecture is becoming more important as AI agents take on larger roles in software development, IT operations and business processes.

“Enterprise architecture is becoming more important, not less,” the analysts wrote. “But the basis of its importance is changing.”

AI is already capable of generating architecture diagrams, drafting standards, documenting systems, analyzing dependencies and summarizing technology portfolios. Tasks that previously required weeks of manual work can increasingly be completed in hours, although Forrester cautions that the quality of AI-generated work remains uneven.

That shift changes where architects can provide value.

If architectural artifacts become inexpensive and abundant, Forrester argues, enterprise knowledge and judgment become the scarcer resources. Architects may increasingly serve as curators of enterprise context, stewards of architectural knowledge and designers of governance mechanisms.

The change becomes particularly important as autonomous AI agents make more decisions. Organizations will need to establish what those systems are authorized to do, what constraints apply and how those constraints are enforced.

Forrester describes this emerging role as a “control plane for bounded autonomy.”

The analysts also envision architects helping organizations create reusable enterprise intelligence layers containing policies, standards, decision records, dependencies, business context and other information that can be consumed by both humans and AI systems.

Architecture itself could consequently become more continuous. Instead of relying primarily on periodic architecture reviews, AI could allow architectural guidance and governance to become embedded directly into technology delivery workflows.

The article accompanies Forrester’s new report, “The AI Enterprise Architect,” which examines which architecture responsibilities are likely to be automated and which could become more important.

The larger argument is that AI may threaten some of the work enterprise architects perform without eliminating the need for architecture itself. If Forrester is correct, the defining question for the profession may shift from who produces architectural artifacts to who establishes the knowledge, authority and guardrails that increasingly autonomous systems use to make decisions.

We reached out to a handful of experts – Dr. Alok Mehta,  Master Scientist Leonard Greski, Dr. Magesh Kasthuri, Dr. Tushar Hazra, Daniel Lambert, M. Sc. – to get their opinions to two questions:

Question: Which parts of an enterprise architect’s job are most vulnerable to AI?

Mehta: I agree that AI will automate much of the artifact-producing work of enterprise architecture, including documentation, diagrams, inventories, technology research, initial solution options and standards. This is positive because those activities are not where experienced architects provide their greatest value. An architect’s real value lies in judgment: understanding the business, challenging unnecessary complexity, anticipating downstream consequences and balancing security, data, integration, resiliency, cost and operational risk. AI can accelerate those decisions, but it does not replace the experience and organizational context required to make them responsibly.

Greski: The fact that AI can automate the artifact-producing work of enterprise architecture is a symptom, not the root problem.  The existential threat to Enterprise Architecture is a lack of measurable results, not AI. I specifically addressed the absence of results problem in my 2022 A&G Magazine article, Getting Beyond Ourselves.  The profession has made little progress on results oriented architecture since the article’s publication.

Kasthuri: Based on my experience, some activities like review, design optimization, design strategy, cost implications, identifying risks and mitigation plans, design traceability, architectural decisions can be automated through AI but the success lies in how much can you believe the outcome of AI generated information for the above tasks. It means that you can use AI but can you believe AI for these tasks. There should be thorough audit and verification process to enable these automated agents and it cannot happen without HITL.

Hazra: There is a lot of truth in this argument. I know architecture documentation leads to a slew of activities that an enterprise architect performs. So verifying and validating AI-prepared documentation is essential. I have faced many barriers to applying due diligence, often to the point that it hinders an initiative’s progress. I argue that it ultimately pays off. Validated documentation allows enterprise architects to align business needs with technical solution options, select appropriatI assert that enterprise architects add value through standards and compliance reviews, architectural traceability and impact analysis, market outlook and vendor research, and presenting final assessments and reports.e technology, tools, and techniques, and, most importantly, maintain a regular cadence of governance for business and technology alignment and enablement.

Lambert: I agree with Forrester’s central argument. AI is unlikely to eliminate enterprise architecture, but it will substantially change where enterprise architects create value. The activities most vulnerable are those involving repetitive analysis, documentation, repository maintenance, diagramming, standards generation, dependency analysis, and portfolio assessment. Forrester notes that work that historically took weeks can increasingly be performed in hours. The architect’s role shifts toward enterprise context, judgment, governance, decision rights, accountability, and managing trade-offs.

This aligns closely with Forrester’s argument that architecture artifacts were never the ultimate objective; they demonstrated an architect’s understanding of the enterprise. As AI makes artifact production inexpensive and fast, that enterprise understanding becomes the scarce asset. This transition is explored in this article, “Building AI-Enabled Enterprise Architecture Workflows,” which examines how AI can transform traditional EA activities into intelligent, automated workflows that accelerate analysis, generate and maintain architecture artifacts, and enable architects to focus more on strategic insight, governance, and decision-making.