All cases AI Strategy & Adoption

Building a virtual enterprise architect

Enterprise architecture only delivers its full value when people can access and interpret it. The diagrams, relationships, and decisions captured in an architecture model and repository represent years of organizational knowledge, but in most organizations that knowledge sits behind a single point of access: the architect. Every question, every impact analysis, every onboarding request flows through one person.

That bottleneck is not a people problem. It is an interface problem.

The challenge

At iChoosr, a growth-stage technology company, a well-maintained ArchiMate architecture model covered the full organizational landscape, from business processes and applications through to data and technology infrastructure. The model was genuinely valuable, but accessing that value required ArchiMate experience and constant architect involvement. An auditor needing an IT infrastructure overview, a business team exploring a new market, a new team member trying to understand the platform, a developer checking whether their implementation aligned with the architecture. Each request landed with the architect, required context switching, and took time.

Building the virtual architect agent

I designed and built an AI agent with direct, real-time access to the sources that matter: the ArchiMate architecture model, the project and documentation tooling, the code repository, and security standards. Using MCP integrations, the agent could query all of these simultaneously and synthesize responses spanning the business, application, data, and technology layers.

The agent does not replace architectural judgment. It handles the analytical groundwork so that judgment can be applied where it counts.

What it delivered

Three use cases demonstrated the value most clearly.

Audit support

The agent queried the architecture model and existing documentation, synthesized the findings across the relevant domains, and produced structured output ready for human review. What had been a multi-hour task of manual compilation became a starting point requiring only validation and final polish.

Implementation governance

The agent performed a full architecture-to-code alignment analysis, comparing the documented application layer against the live codebase. It identified which components were correctly implemented, flagged a critical gap between documentation and reality, and surfaced several architectural deviations for review, all in minutes rather than days.

Strategic impact analysis

The agent assessed the business, application, and technology implications of expanding into a new market, tracing dependencies across the full architecture model to surface considerations that would otherwise require significant manual effort to compile.

The broader point

An architecture model maintained with discipline is a strategic asset well beyond its traditional audience. Paired with an AI agent, it becomes accessible to every stakeholder who needs architectural insight, without creating a bottleneck. The architect's role shifts from being the interface to the knowledge, to being the person who ensures the knowledge stays current, the outputs stay accurate, and the decisions stay sound.

This is the kind of practical AI application grounded in real organizational context, connected to the systems that already hold the answers, and designed to amplify the people doing the work.

Let's Talk

Interested in a similar approach?

Whether you're looking to unlock the value of existing knowledge assets or exploring practical AI applications for your organization, let's explore how we can work together.