Methodology
Phase 3 of 4

Operation

Manage what was delivered

What Happens in Operation

Operation is where delivered work meets the real world. Completed units are deployed, monitored, and maintained. This phase ensures that what was built continues to function as intended.

Operation is not an afterthought. It is a first-class phase with its own structured activities, quality gates, and collaboration modes.

Key Activities

  • Deploy or publish: Move completed work from development to production. In software, this means deployment. In other domains, it means publishing, distributing, or activating deliverables.
  • Monitor outcomes: Track whether the work achieves what the completion criteria defined. Are the metrics moving? Are users behaving as expected?
  • Respond to issues: When problems arise, triage and address them. Not every issue requires a full execution cycle — some are operational fixes that can be handled in-place.
  • Maintain state: Keep documentation, configurations, and operational artifacts current. Drift between documentation and reality is a form of technical debt.

How the AI Agent Behaves

During operation, the agent shifts to a monitoring and maintenance role. It can perform routine operational tasks, generate reports, and flag anomalies.

For more complex operational needs — incident response, significant configuration changes, or architectural modifications — the agent escalates to the team or initiates a new elaboration cycle.

Across Domains

| Domain | Operation Looks Like | |---|---| | Software | Deployment, monitoring, incident response, dependency updates, security patches | | Marketing | Campaign launch, performance tracking, A/B test monitoring, content updates | | Operations | Go-live, SLA monitoring, vendor management, compliance tracking | | Research | Publication, citation tracking, methodology documentation, data archival |

Outputs

  • Deployment records: What was deployed, when, and to where
  • Operational metrics: Performance data tied back to completion criteria
  • Incident logs: Issues encountered and how they were resolved
  • Maintenance artifacts: Updated documentation, runbooks, and configurations