The COMPEL Transformation Atlas
Implementation Roadmap
A single, navigable map of how to adopt COMPEL — six stages across three transformation enablers, with every artifact, control, role, and credential linked to its source. Designed to be read in ten seconds and explored for an hour.
A continuous transformation ribbon: six COMPEL stages, from Go COMPEL through to Business Outcomes. Cert pins mark where training unlocks each leg of the journey.
- Value thesis & baseline
- 18-domain maturity scan
- Agent inventory & autonomy class
- Portfolio & funding model
- Operating model & RACI
- Agent policy & approval boundaries
- Outcome blueprints
- Target operating model & platform
- Agent control design
- Compound, retire, re-thesis
- Capability uplift & playbook refresh
- Agent policy evolution
- 30 / 60 / 90 benefit reviews
- Control effectiveness & residual risk
- Agent audit & adversarial testing
- Value telemetry in flight
- MLOps, monitoring, runbooks
- Agent runtime controls live
& Grow
- Value · Realized ROI
- Readiness · Control effectiveness
- Agents · Audited autonomy
Calibrate
See clearly before you commit
Value Value thesis & baseline
Translate corporate strategy into testable AI value theses with measurable baselines and a pipeline of candidate use cases.
Readiness 18-domain maturity scan
Score the 10 readiness dimensions and the 18-domain maturity model to find the real capability gaps.
Agents Agent inventory & autonomy class
Discover every autonomous agent in the organization and classify by autonomy level and risk tier.
→ Honest baseline of AI maturity, risk, and value potential
Organize
Mobilize the operating system
Value Portfolio & funding model
Stand up the value-funded portfolio, KPI tree, funding guardrails, and benefit-tracking cadence.
Readiness Operating model & RACI
Charter the CoE, define decision rights, and stand up the named roles that make AI sustainable.
Agents Agent policy & approval boundaries
Establish the agent policy defining tool access, data classification, and approval thresholds.
→ Sponsored governance, funded portfolio, named accountabilities
Model
Architect the target state
Value Outcome blueprints
Translate value theses into outcome blueprints, target KPI trees, and attribution models.
Readiness Target operating model & platform
Design the target operating model, control library, and ML/data platform reference architecture.
Agents Agent control design
Design HITL thresholds, kill-switch mechanisms, escalation paths, and simulation test plans.
→ Approved target operating model, controls, and platform blueprint
Produce
Build and deploy with discipline
Value Value telemetry in flight
Instrument every system at deployment so KPIs flow into the value dashboard from day one.
Readiness MLOps, monitoring, runbooks
Build the production pipeline, drift monitors, incident runbooks, and on-call rotation.
Agents Agent runtime controls live
Activate tool-access controls, decision audit trail, and graceful-shutdown mechanisms in production.
→ Production AI systems with evidence trails and value telemetry
Evaluate
Measure what was promised
Value 30 / 60 / 90 benefit reviews
Run the post-deployment review cadence and reconcile delivered value against the original thesis.
Readiness Control effectiveness & residual risk
Test control effectiveness, refresh readiness scores, and record residual risk on the register.
Agents Agent audit & adversarial testing
Audit decision logs, run adversarial and chaos tests, and refresh the agent risk tier.
→ Realized value, validated controls, residual risk on the record
Learn
Compound the gains, retire the losses
Value Compound, retire, re-thesis
Scale what worked, retire what didn't, and feed the next cycle's value thesis.
Readiness Capability uplift & playbook refresh
Update the playbook, refresh certifications across the workforce, and re-baseline maturity.
Agents Agent policy evolution
Tune autonomy levels, update escalation rules, and evolve the agent policy from operational evidence.
→ Updated playbooks, capability uplift, next-cycle thesis