01
Workflow mapping
Map current-state workflow, decision points, exception paths, data sources, and ownership. Output: workflow map and friction inventory.
Framework 04
Everything RSUA does, mapped against value path, Human Workbench design, implementation, and governed expansion.

RSUA work follows the four phases of the Process Framework: Define, Convert, Install, and Govern. Each capability is named, scoped, and tied to an owner, output, value path, Human Workbench, and measurement gate. No vague retainers. No scope creep without explicit re-scoping.
01
Map current-state workflow, decision points, exception paths, data sources, and ownership. Output: workflow map and friction inventory.
02
Score the operating environment across data, process, ownership, and governance. Output: readiness score, blockers, gates, and first-workflow evidence plan.
03
Rank candidate AI-native workflows by value, feasibility, and risk. Output: first-workflow sequence, value lever, ROI test plan, and evidence needed before build.
01
Choose the AI pattern, tool stack, and integration approach for the converted workflow. Output: phase-one system spec with build, buy, or partner calls per component.
02
Define confidence thresholds, escalation paths, reviewer roles, audit requirements, correction capture, and rollback boundaries. Output: human-role and control plan.
03
Specify what the accountable person sees before deciding: evidence, confidence, risk, validation errors, allowed actions, reviewer burden, and what each decision does next. Output: Human Workbench spec.
04
Translate the AI workflow into measurable business metrics with named owners. Output: value model, payback hypothesis, and gating criteria.
01
Implementation of agents, tools, and integrations against the converted workflow. Output: deployed system with evals and acceptance gates.
02
All five layers: input, retrieval, generation, action, audit. Output: documented control set and red-team report.
03
Run the system in production parallel without affecting live operations. Output: accepted-output report, reliability readout, and edge-case inventory.
01
Recurring eval runs, drift detection, and threshold tuning. Output: monitoring cadence and performance readout.
02
Reduce human review on workflow segments where reliability has been demonstrated. Output: updated control matrix.
03
Add new tools, data sources, or adjacent workflows to the existing system. Output: incremental release notes and updated runbook.

Continue Reading
Framework 05