How XPlain, AAMAS, and STAR XAI Fit Together
XPlain is the governance-and-assurance layer for real-world AI; IFAAMAS supplies the agentic constructs and coordination norms; XAI provides interpretability of model and decision behavior. Together they cover what is permitted, how agents act, and why results occur.
Legend: ✓ = primary role · ↔ = complements/joins · — = not a focus
| Concern / Layer | XPlain (Governance & Assurance) | AAMAS (Agentic Constructs) | XAI (Interpretability) |
|---|---|---|---|
| Purpose | ✓ Define rules, evidence gates, pause points, assurance checks | ✓ Define agents’ goals/commitments, norms, protocols | ✓ Explain model/decision behavior to humans |
| Goals & Commitments | ↔ Encodes objectives and thresholds in Primers | ✓ BDI (Beliefs–Desires–Intentions), institutional norms | ↔ Explains chosen goals/plan rationale |
| Reasoning & Execution | ✓ Romer governs lenses, rules, escalation, and trace | ✓ Agent plans, interactions, coordination patterns | ↔ Provide step/rule/model explanations |
| Coordination & Protocols | ↔ Process handoffs and audited interactions | ✓ Protocols (e.g., contract-net, norms, messaging) | — Not a protocol framework |
| Human-in-the-Loop | ✓ Mandatory review triggers (confidence, boundary, ethics) | ↔ Escalation/commitment rules integrate with HiTL | ✓ Surfaces reasons so reviewers can act |
| Assurance & Audit | ✓ Confidence floors, drift checks, Romer Trace, reports | ↔ Verifiable commitments, compliance with norms | ✓ Evidence-backed explanations (local/global) |
| Data & Evidence | ✓ Evidence gates, provenance, change budgets | ↔ Beliefs can reference governed evidence | ✓ Shows which inputs/features influenced results |
| Model Runtime / Vendor | ✓ Vendor/runtime-agnostic governance wrapper | ↔ Conceptual layer; can sit over many runtimes | ↔ Techniques span many model types |
| Primary Output | Assured decision + audit trace | Coordinated agent behavior | Human-readable explanations |
| Non-Goals | — Not a model or protocol spec | — Not an assurance or XAI toolkit | — Not a governance or agency framework |
Takeaway: XPlain governs and proves what should happen, AAMAS structures how agentic behavior unfolds and coordinates, and XAI explains why each result is produced—so you can scale from API calls to full agent ecosystems without losing accountability.