COMPEL Specialization Stream · AITE-ATS
COMPEL Academy — Agentic AI Systems Architect Expert
Advanced expert certification for practitioners designing and governing agentic AI systems: autonomy classification, tool controls, human-in-the-loop design, and agentic evaluation metrics.
Profession title: Agentic AI Systems Architect Expert
Audience: Governance leads, architects, and senior practitioners deploying agentic AI systems in production.
Enroll in the AITE-ATS track
Registration, enablement, and the proctored assessment are delivered through compel.one. Seats open continuously.
Prerequisite chain
- AITF AI Transformation Foundations
- → AITE-ATS (this credential)
Learning outcomes
The learning journey is sequenced to cover each outcome below in order. Every article in the journey maps to at least one outcome.
- 1. Classify agent autonomy and design matched control envelopes.
- 2. Govern tool access and design safe escalation paths.
- 3. Instrument agentic evaluation metrics that satisfy auditors.
Body of Knowledge articles (53)
Module M1.2 (53 items)
- Article What Is an Agentic AI SystemM1.2-Art01
- Article Autonomy Spectrum and Agent TaxonomyM1.2-Art02
- Article Agent Runtimes and Execution SubstratesM1.2-Art03
- Article Agent Loop Patterns — ReAct, Plan-and-Execute, Reflexion, State-GraphM1.2-Art04
- Article Tool Use — Schemas, Registries, and Design DisciplineM1.2-Art05
- Article Tool-Call Authorization and ValidationM1.2-Art06
- Article Memory Architecture for AgentsM1.2-Art07
- Article Goal Hijacking, Excessive Agency, and Prompt-Injection CascadesM1.2-Art08
- Article Kill-Switch Architecture and Escalation ProtocolsM1.2-Art09
- Article Human-in-the-Loop and Human-on-the-Loop DesignsM1.2-Art10
- Article Multi-Agent Orchestration — Framework ComparisonM1.2-Art11
- Article Agent-to-Agent Communication and Coordination FailuresM1.2-Art12
- Article Agentic RAG and Dynamic Knowledge AccessM1.2-Art13
- Article Indirect Prompt Injection and Supply-Chain AttacksM1.2-Art14
- Article Observability for Agentic SystemsM1.2-Art15
- Article Operational Resilience for Agents — Failure Modes and RecoveryM1.2-Art16
- Article Agent Evaluation and Simulation HarnessM1.2-Art17
- Article Agent SLO/SLI and Operational MetricsM1.2-Art18
- Article Cost Architecture for Agentic WorkloadsM1.2-Art19
- Article Agentic Platform DesignM1.2-Art20
- Article Sandboxing and Execution Isolation for AgentsM1.2-Art21
- Article Policy Engines for Agentic Action GatingM1.2-Art22
- Article EU AI Act Articles 14, 52, and Conformity Assessment for Agentic SystemsM1.2-Art23
- Article Agent Lifecycle, Versioning, and PromotionM1.2-Art24
- Article Incident Response for AgentsM1.2-Art25
- Article Agent, Prompt, Tool, and Memory RegistriesM1.2-Art26
- Article Security Architecture for Agentic SystemsM1.2-Art27
- Article Data Architecture for Agentic SystemsM1.2-Art28
- Article Multi-Agent Patterns: Hierarchical, Market, Swarm, ActorM1.2-Art29
- Article Financial Services Agentic PatternsM1.2-Art30
- Article Healthcare and Life Sciences Agentic PatternsM1.2-Art31
- Article Public-Sector Agentic PatternsM1.2-Art32
- Article Software-Engineering Agentic PatternsM1.2-Art33
- Article Responsible-Agentic-AI Pattern LanguageM1.2-Art34
- Article Operating Model for Agentic SystemsM1.2-Art35
- Article Architect in Calibrate and Organize Stages for Agentic SystemsM1.2-Art36
- Article Architect in Model and Produce Stages for Agentic SystemsM1.2-Art37
- Article Architect in Evaluate and Learn Stages for Agentic SystemsM1.2-Art38
- Article Build vs. Buy for Agentic Platform ComponentsM1.2-Art39
- Article Capstone — A Complete Agentic Reference Architecture PackageM1.2-Art40
- Lab Lab — Build a Finance Agent with a Human-in-the-Loop Escalation MatrixM1.2-Art51
- Lab Lab — Design a Tool-Use Guardrail Matrix for a Coding AgentM1.2-Art52
- Lab Lab — Build an Agent Observability Dashboard and Session ReplayM1.2-Art53
- Lab Lab — Red-Team an Agent for Indirect Prompt InjectionM1.2-Art54
- Lab Lab — Design an Agent Kill-Switch Specification with Escalation ProtocolsM1.2-Art55
- Case Study Case Study — Moffatt v. Air Canada as an Agentic-Governance Failure, from the Architect's SeatM1.2-Art61
- Case Study Case Study — Devin and the Replit Agent: Coding-Agent Incidents from the Architect's SeatM1.2-Art62
- Case Study Case Study — Anthropic Computer Use as a Controlled-Rollout ArchitectureM1.2-Art63
- Template Template — Agent Governance Charter (AITE-ATS instantiable per-agent)M1.2-Art71
- Template Template — Tool-Use Constraint SpecificationM1.2-Art72
- Template Template — Agent SLO / SLI SheetM1.2-Art73
- Template Template — Escalation Matrix (HITL / HOTL / Autonomous)M1.2-Art74
- Template Template — Kill-Switch RunbookM1.2-Art75
Competencies demonstrated
- → Agent autonomy classification and controls
- → Tool access governance and safety
- → Human-in-the-loop and escalation design
- → Agentic evaluation metrics and monitoring
Exam blueprint summary
- Assessment
- Proctored examination
- Passing score
- 80% passing score
- Portfolio
- Required
- Renewal
- Every 24 months
- Recommended hours
- 80
- CE credits
- 80
Linked Core Mastery context
The Specialization Stream assumes AITF Foundations fluency. These Core Mastery resources are the recommended grounding before entering the AITE-ATS learning journey.
Formal credential definition
The machine-readable Open Badges 3.0 / W3C Verifiable Credential
definition for AITE-ATS is published at
/credential/aite-agentic-transformation
. HR platforms and AI citation engines can fetch the JSON-LD
document at
/credential/aite-agentic-transformation.json
.