Rearchitecting How We Work

The most transformative change in enterprise software isn’t new applications—it’s realizing we need fewer humans to operate our existing ones. After decades of digitization and cloud migration, we’re entering an era where software actively participates in the work itself.

This shift extends far beyond productivity gains, impacting architecture, economics, security, governance, and even how we define the enterprise operating model. While traditional application strategy focused on functionality, integration, and ROI, a new question emerges: “Who or what should perform this work?”

The Rise of Agent Ecosystems

Gartner predicts that by 2026, 40% of enterprise applications will feature task-specific AI agents—a significant jump from less than 5% today. These agent ecosystems are spanning functions beyond individual apps.

I call this emerging discipline Enterprise Work Architecture (EWA): the deliberate design of how an outcome moves across human judgment, machine intelligence, data, and authority—from intent to execution and evidence.

From Systems-Centric to Outcome-Driven

Traditional architecture organizes capabilities around discrete systems like ERP, CRM, and HCM. Business outcomes rarely respect these boundaries, creating friction as humans navigate between them.

AI agents compress these seams by gathering context, interpreting intent, and coordinating actions—redesigning how work travels across applications rather than simply adding AI within apps.

The Six Steps of EWA

  1. Outcome: What measurable result are we trying to achieve?
  2. Work: Which activities are necessary, which can disappear, and where should human judgment reside?
  3. Authority: What may a person, application, or agent recommend, prepare, approve, or execute?
  4. Execution: Through which systems, APIs, data, and workflows must this travel?
  5. Evidence: Can consequential actions be observed, reconstructed, challenged, and recovered?
  6. Economics: How does this redesign impact time, cost, quality, risk, and experience?

The sequence matters—EWA starts with the desired outcome and designs backward.

Example: Resolving Customer Billing Disputes

Today’s process often requires employees to navigate across multiple systems before a credit can be approved. With EWA:

  • Agents gather and reconcile data automatically
  • Humans handle exceptions requiring nuanced judgment
  • Clear authority levels dictate what actions machines can take
  • All steps are traceable for auditability and continuous improvement

This is the difference between automating tasks and rearchitecting entire workflows—unlocking true productivity potential while preserving human oversight.