On 18 September, Forrester analysts Charles Betz and Joseph Schiavone published a piece with a deliberately blunt title: "Will AI Eliminate Enterprise Architects?" Their answer was no, but not for the reason most architects would hope. "Enterprise architecture is becoming more important, not less," they wrote. What's changing is why it matters.
The piece accompanies a new Forrester report, "The AI Enterprise Architect," and its argument lands at an awkward moment for many EA teams. Much of what those teams spend their time producing can now be generated in a fraction of the time.
The Work That Is Being Automated
Forrester is direct about this. AI can already generate architecture diagrams, draft standards, document systems, analyse dependencies and summarise technology portfolios. Work that used to take weeks can increasingly be done in hours. Forrester notes that the quality is still uneven, and anyone who has reviewed an AI-generated capability map knows that caveat is real. But the direction is clear.
If an EA team's value is measured by the artefacts it produces, that value is shrinking. The diagrams were never really the point. They were evidence that someone understood how the enterprise actually works.
What Becomes Scarce Instead
Forrester's argument is that when artefacts become cheap and plentiful, enterprise knowledge and judgement become the scarce resources. Architects shift toward curating enterprise context, stewarding architectural knowledge and designing the governance mechanisms that autonomous systems operate within.
The phrase Forrester uses for this emerging role is a "control plane for bounded autonomy." As AI agents make more decisions, someone has to define what those agents are authorised to do, what constraints apply and how those constraints are enforced. That's an architecture problem, and it's arguably the most consequential one enterprises face right now.
From Periodic Review to Continuous Architecture
The second shift is structural. Architecture review boards that meet monthly were designed for a world where change moved at the pace of projects. Agents don't wait for the next board meeting. Forrester suggests architectural guidance and governance will increasingly be built directly into delivery workflows rather than applied at review checkpoints.
It also describes a "reusable enterprise intelligence layer": policies, standards, decision records, dependencies and business context stored so that both people and AI systems can use them. In practice, that means architecture knowledge has to be machine-readable, not locked in slide decks and wiki pages nobody updates.
What an EA Team Should Actually Change
For most teams, the practical shift over the next year looks something like this. Stop measuring the function by artefacts produced and start measuring it by decisions enabled and risks avoided. Move standards and policies into formats that agents and delivery pipelines can actually read. Define decision rights for autonomous systems explicitly, including where human approval stays mandatory. And use AI to automate the documentation work, so architects spend their time on the judgement calls it can't make.
What This Means for Your Organisation
What we see across the enterprise architecture engagements we run is that the EA teams under the most pressure are the ones that became documentation functions. They're easy to question when AI can produce a first draft of their output in an afternoon. The teams in the strongest position already act as the organisation's source of context and decision rights. Forrester's analysis doesn't threaten those teams. It describes the job they're already growing into.
Key Takeaways
- Forrester's 18 September analysis argues AI will not eliminate enterprise architects, but will change the basis of their value from producing artefacts to supplying judgement and context.
- AI can already generate diagrams, standards, documentation and dependency analysis in hours rather than weeks, although quality remains uneven.
- Forrester describes an emerging EA role as a "control plane for bounded autonomy", defining what AI agents may do and how constraints are enforced.
- EA teams should shift toward machine-readable standards, explicit decision rights for autonomous systems and governance built into delivery workflows rather than periodic review.
How Trusenta Can Help
Enterprise Architecture in the AI Era helps EA teams move from producing artefacts to defining decision rights and constraints for autonomous systems.
Enterprise Architecture keeps architecture knowledge, decisions and dependencies in a structured form that both people and AI systems can use.
Fractional Enterprise Architect gives organisations without a mature EA function the judgement and governance design Forrester argues is now the scarce resource.
Conclusion
The question Forrester posed was designed to provoke, and the answer is more useful than a simple yes or no. Enterprise architecture isn't going away. The version of it built around diagrams and review boards probably is. The EA teams that move now toward context, decision rights and continuous governance will find their role becoming more central as autonomous systems spread, not less.
