How to run an architecture time allocation baseline with your team

Before an architecture team changes how it works, it helps to know how it spends its time now. The real allocation across the work that matters can differ from the job description, from what leadership assumes, and from what the team would guess. Measuring it is the first step in moving from manual work to AI-Augmented Architecture, because it shows where the mechanical effort sits and where AI has something to take on.

The result is some of the most useful data a practice can hold. It shows where the productivity drag sits, which improvement would make the most difference, and where AI augmentation has hours to recover. It also turns the leadership conversation from opinion into evidence. You do not need anyone to run this for you, and it takes about half a day.

The four domains

Architecture work sorts into four categories, and the frame holds for a team of two or a team of twenty. For each, here is what the work looks like by hand today and what changes when it is AI-augmented.

Architecture Modeling is putting information into the repository: interviews and workshops to gather what is true, creating and editing elements in Sparx EA, diagramming, writing descriptions and metadata, and wiring up relationships. AI-augmented, the AI drafts the content from source material and the architect reviews and decides.

Architecture Analysis is getting insight back out of the model: impact analysis (if I change this application, what breaks?), dependency mapping, compliance reporting, cross-system queries, and pulling data for a planning decision. AI-augmented, the AI compiles the answer from the model and the architect interprets it.

Architecture Governance is keeping the repository trustworthy and the team working to standard: standards checking, review-board preparation, completeness validation, MDG Technology conformance, logging decisions and capturing lessons learned. AI-augmented, the AI runs the checks against rules the architect wrote, and people handle the exceptions.

Stakeholder Engagement is fielding questions from and briefing people who are not architects: emails about who owns a system, current-state diagrams for a planning session, explaining why a decision went the way it did, and meetings spent answering questions the model should already answer. AI-augmented, routine questions are answered from the model, and the architect’s time goes to conversations that change minds.

Between them, these four absorb nearly all of a team’s time, with a small remainder for admin, training and work that fits nowhere.

The exercise, step by step

How to read the result

Look first at how much of the week goes to the mechanical parts of modeling and governance: transcribing content, laying out diagrams, checking attributes. In a manual practice that share is usually large, and analysis, the work that drives strategy, gets squeezed. That pattern is what AI-Augmented Architecture is built to change.

As AI carries the mechanical load, the profile should shift. Modeling time falls because AI drafts the content and architects review it. Governance time falls because checking runs automatically and people handle the exceptions. Analysis time can rise, because each question costs less to answer. Engagement time changes character, with fewer “who owns this?” interruptions and more conversations that change decisions.

Some variation is healthy. A governance-heavy team may work in a high-regulation environment, and an engagement-heavy one may support a large, distributed organization. The real question is alignment: does this allocation match your priorities? When half the effort goes to answering email while analysis is what drives strategy, you have something to fix.

The skill this asks of a leader is reading the numbers without blame, and choosing which domain to change first.

What to do with the baseline

Repeat the measure quarterly. Annually is too slow, because the work changes faster than that. As you introduce AI-augmented work, watch whether governance time comes back, and whether stakeholder self-service shrinks the engagement domain or only changes its texture.

Keep the worksheet you used, with the four domains and your team’s split. Run it again in two quarters and you will have evidence about whether your changes are landing. Measuring the baseline, naming the target and holding the team to moving toward it is how that accountability starts. For a closer look at the four domains, see where architect time goes, and how AI-augmented work changes it, or start from where your architects spend their day.

If you want help turning the baseline into a plan, Paralysis to a Plan works through the problem, the pieces, what success looks like and the plan with your team.