{"id":304,"date":"2025-11-20T14:06:54","date_gmt":"2025-11-20T14:06:54","guid":{"rendered":"https:\/\/xplain-r.com\/?page_id=304"},"modified":"2026-05-07T12:13:01","modified_gmt":"2026-05-07T12:13:01","slug":"architecture","status":"publish","type":"page","link":"https:\/\/xplain-r.com\/fr\/about\/architecture\/","title":{"rendered":"Architecture"},"content":{"rendered":"
\n

XPlain is a general-purpose form of agentic AI that\u2019s platform-agnostic and audit-ready. It orchestrates decisions using Meta-Primers, domain Primers, and Romers to align beliefs, goals, and commitments with evidence gates, human pause points, and assurance controls. The result: agents you can trust\u2014traceable, compliant, and ready for real-world deployment across GRC, supply chain, risk, and data quality.<\/p>\n<\/div>\n\n\n\n

\u201cMeta-primer governs both the topical primers and the romer, ensuring cross-aligned behaviour and trace policy\u201d<\/em><\/p>\n\n\n\n

“XPlain does the how, not the what”<\/em><\/p>\n\n\n\n

\n

XPlain-R Romer Trace \u2013 Full Architecture Overview<\/h3>\n\n

\n The Romer Trace function sits inside the XPlain-R framework as the audit and assurance spine.\n It connects the meta-primer (global rules), topical primers (job-specific rules), and the romer\n runtime (navigation engine) into an explainable, versioned trace that can be reviewed and compared\n across platforms and time.\n <\/p>\n\n

Layers<\/h4>\n
    \n
  • Meta-Primer Layer<\/strong> \u2013 defines global reasoning constraints, logging obligations, and governance rules that apply to every job.<\/li>\n
  • Topical Primer Layer<\/strong> \u2013 binds the meta-primer to a specific domain or use case (e.g. GRC assessment, supply chain audit).<\/li>\n
  • Romer Runtime Layer<\/strong> \u2013 executes the combined rules, generates waypoints and sub-questions, and drives the reasoning path.<\/li>\n
  • Trace Layer<\/strong> \u2013 normalises and persists romer events as an ordered, version-aware trace with state snapshots.<\/li>\n
  • Storage & Analytics Layer<\/strong> \u2013 provides immutable trace storage and cross-job analytics on romer behaviour and method performance.<\/li>\n
  • Experience Layer<\/strong> \u2013 exposes traces to humans via viewers, dashboards, and export packs for audit and research.<\/li>\n <\/ul>\n\n

    Key Design Properties<\/h4>\n
      \n
    • Version-aware<\/strong> \u2013 every trace event is tagged with primer and meta-primer versions, so behaviour can be explained in context.<\/li>\n
    • Append-only<\/strong> \u2013 traces are immutable; any corrections are new events, not edits.<\/li>\n
    • Separation of concerns<\/strong> \u2013 primers define rules<\/em>, the romer defines navigation<\/em>, and the trace service defines auditability<\/em>.<\/li>\n
    • Cross-platform<\/strong> \u2013 the event schema can be implemented across different AI backends while preserving a common trace format.<\/li>\n
    • Assurance-ready<\/strong> \u2013 designed to support architecture review, internal and external audit, and regulatory explainability requirements.<\/li>\n <\/ul>\n\n

      Typical Flow<\/h4>\n
        \n
      1. A job is started with a meta-primer and a topical primer bound to it.<\/li>\n
      2. The romer runtime applies these rules to explore, evaluate, and synthesise, emitting structured events.<\/li>\n
      3. The Romer Trace Service enriches these events with versions and state snapshots and writes them to the Trace Store.<\/li>\n
      4. Analytics and observability tools read from the Trace Store to surface patterns, anomalies, and performance metrics.<\/li>\n
      5. Trace viewers and export packs present a human-readable reconstruction of what happened and why.<\/li>\n <\/ol>\n<\/section>\n\n\n\n
        \n\n

        Romer Trace Architecture<\/h2>\n\n

        \n High-level component and data-flow view for the Romer Trace function.\n <\/p>\n\n \n

        \n\n \n
        Meta-Primer<\/div>\n
        Global Rules \u00b7 Governance \u00b7 Versioning<\/div>\n
          \n
        • \u2022 Global reasoning constraints<\/li>\n
        • \u2022 Event logging & trace policies<\/li>\n
        • \u2022 Meta-primer version & IDs<\/li>\n <\/ul>\n <\/div>\n\n \n
          Pr\u00e9misse de base<\/div>\n
          Job-Specific Guidance<\/div>\n
            \n
          • \u2022 Contextual rules for this job<\/li>\n
          • \u2022 Domain patterns & templates<\/li>\n
          • \u2022 Primer version & IDs<\/li>\n <\/ul>\n <\/div>\n <\/div>\n\n \n
            \n Meta-Primer<\/span>\n \u27f7<\/span>\n Pr\u00e9misse de base<\/span>\n (alignment, constraints & overrides)<\/span>\n <\/div>\n\n
            \n
            \u2b07<\/div>\n
            combined rules + versions<\/div>\n <\/div>\n\n \n
            \n \n
            \n Source of interpretive navigation\n <\/div>\n
            Romer Runtime<\/div>\n
            Navigation \u00b7 Modes \u00b7 Waypoints<\/div>\n
              \n
            • \u2022 Executes meta-primer constraints & topical rules<\/li>\n
            • \u2022 Generates waypoints & sub-questions<\/li>\n
            • \u2022 Manages modes (explore \/ evaluate \/ synthesize \/ check)<\/li>\n
            • \u2022 Emits events (JobStarted, RomerWaypoint, EvidenceAttached, ExternalCheckRecorded, OutcomeCommitted, JobCompleted)<\/li>\n <\/ul>\n <\/div>\n <\/div>\n\n
              \n
              \u2b07<\/div>\n
              event stream<\/div>\n <\/div>\n\n \n
              \n \n
              \n Normalisation & Governance Layer\n <\/div>\n
              Romer Trace Service<\/div>\n
              Event Enrichment & Persistence<\/div>\n
                \n
              • \u2022 Normalises event payloads<\/li>\n
              • \u2022 Attaches primer & meta-primer versions<\/li>\n
              • \u2022 Captures romer state (mode, focus, constraints)<\/li>\n
              • \u2022 Maintains ordered, append-only trace per job<\/li>\n
              • \u2022 Applies privacy filters & access rules<\/li>\n <\/ul>\n <\/div>\n <\/div>\n\n
                \n
                \u2b07<\/div>\n
                append-only trace records<\/div>\n <\/div>\n\n \n
                \n\n \n
                Trace Store<\/div>\n
                Immutable Audit Ledger<\/div>\n
                  \n
                • \u2022 Per-job event streams<\/li>\n
                • \u2022 Indexed by job, agent, version, time<\/li>\n
                • \u2022 Supports filtered read & export<\/li>\n <\/ul>\n <\/div>\n\n \n
                  Analytics & Observability<\/div>\n
                  Behaviour Across Runs<\/div>\n
                    \n
                  • \u2022 Romer path patterns & clustering<\/li>\n
                  • \u2022 Primer rule usage analytics<\/li>\n
                  • \u2022 Dashboards & health metrics<\/li>\n <\/ul>\n <\/div>\n <\/div>\n\n
                    \n
                    \u2b07<\/div>\n
                    views & dashboards<\/div>\n <\/div>\n\n \n
                    \n\n \n
                    Trace Viewer \/ Job UI<\/div>\n
                    Per-Job Reconstruction<\/div>\n
                      \n
                    • \u2022 Start \u2192 waypoints \u2192 checks \u2192 outcome<\/li>\n
                    • \u2022 Filter by event type or time<\/li>\n
                    • \u2022 Redacted views for external reviewers<\/li>\n <\/ul>\n <\/div>\n\n \n
                      Architects \u00b7 Auditors \u00b7 R&D<\/div>\n
                      Human Consumers<\/div>\n
                        \n
                      • \u2022 Architecture & method review<\/li>\n
                      • \u2022 Assurance & explainability packs<\/li>\n
                      • \u2022 Methodology refinement & experiments<\/li>\n <\/ul>\n <\/div>\n <\/div>\n\n<\/div>\n\n\n\n

                        Flow Summary (End-to-End)<\/strong><\/h3>\n\n\n\n
                          \n
                        1. Meta-Primer & Topical Primer<\/strong>\n
                            \n
                          • Provide rule sets, governance constraints, and version fingerprints.<\/li>\n<\/ul>\n<\/li>\n\n\n\n
                          • Romer Runtime<\/strong>\n
                              \n
                            • Executes the job, generating waypoints and decisions.<\/li>\n\n\n\n
                            • Emits structured events to the Romer Trace Service.<\/li>\n<\/ul>\n<\/li>\n\n\n\n
                            • Romer Trace Service<\/strong>\n
                                \n
                              • Normalises and enriches events.<\/li>\n\n\n\n
                              • Inserts romer state snapshots (mode, focus, constraints).<\/li>\n\n\n\n
                              • Annotates each event with primer + meta-primer versions.<\/li>\n\n\n\n
                              • Writes to an append-only, immutable trace log.<\/li>\n<\/ul>\n<\/li>\n\n\n\n
                              • Trace Store<\/strong>\n
                                  \n
                                • Maintains a fully ordered event stream per job.<\/li>\n\n\n\n
                                • Supports index\/search by job, agent, version, or timeframe.<\/li>\n<\/ul>\n<\/li>\n\n\n\n
                                • Analytics Layer<\/strong>\n
                                    \n
                                  • Extracts path patterns, rule usage profiles, and cross-job behaviour.<\/li>\n\n\n\n
                                  • Powers QA, assurance, and lineage dashboards.<\/li>\n<\/ul>\n<\/li>\n\n\n\n
                                  • Trace Viewer \/ Audit Tools<\/strong>\n
                                      \n
                                    • Presents a job-specific reconstruction:\n
                                        \n
                                      • Start \u2192 waypoints \u2192 checks \u2192 evidence \u2192 outcome.<\/li>\n<\/ul>\n<\/li>\n\n\n\n
                                      • Allows filtered export for architecture and compliance review.<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n\n\n\n
                                        \n

                                        The Stack at a Glance<\/h2>\n\n \n

                                        \n Directive \u2192 Meta-Primer \u2192 Topical Primer \u2192 Romer \u2192 Process \u2192 Job \u2192 Learning\n <\/p>\n\n \n

                                          \n
                                        1. \n Directive<\/strong>\n

                                          Sets ethics, licensing, and scope. Defines permitted artifacts and oversight for auditable operation.<\/p>\n <\/li>\n

                                        2. \n Meta-Primer<\/strong>\n

                                          Construction rules for how domain primers are built, validated, and governed.<\/p>\n <\/li>\n

                                        3. \n Pr\u00e9misse de base<\/strong>\n

                                          Domain logic: evidence gates, thresholds, escalation triggers, and audit requirements.<\/p>\n <\/li>\n

                                        4. \n Romer<\/strong>\n

                                          Executable reasoning derived from the Primer (lens selection, rules, pause points, trace logging).<\/p>\n <\/li>\n

                                        5. \n Process<\/strong>\n

                                          Orchestrates stages and handoffs: intake \u2192 validation \u2192 execution \u2192 review \u2192 outputs \u2192 learning capture.<\/p>\n <\/li>\n

                                        6. \n Job<\/strong>\n

                                          Validated output plus an audit trail suitable for business and system readers.<\/p>\n <\/li>\n

                                        7. \n Learning<\/strong>\n

                                          Feeds improvements back into the Primer (captured edge cases, overrides, drift findings).<\/p>\n <\/li>\n <\/ol>\n<\/section>\n\n\n\n

                                          Why It Works<\/h3>\n\n\n\n
                                            \n
                                          • Scaffold-dependent reasoning:<\/strong> XPlain keeps agents inside explicit boundaries (\u201clenses,\u201d rules, and gates), delivering consistent, auditable results that degrade gracefully on edge cases.<\/li>\n\n\n\n
                                          • Assurance by design:<\/strong> Confidence floors, boundary bands, drift checks, and mandatory pause points are first-class citizens.<\/li>\n\n\n\n
                                          • Full traceability:<\/strong> Every decision can be surfaced via a Romer Trace (inputs \u2192 transformations \u2192 judgments \u2192 outputs).<\/li>\n<\/ul>\n\n\n\n

                                            Layered Architecture (Embed Any Framework, Including IFAAMAS)<\/h3>\n\n\n\n

                                            1) Governance Layer \u2014 Directive<\/strong>
                                            Defines permitted artifacts, interpretive scope, and oversight. This keeps the system auditable and non-opaque across platforms.<\/p>\n\n\n\n

                                            2) Construction Layer \u2014 Meta-Primer & Primers<\/strong><\/p>\n\n\n\n

                                              \n
                                            • Meta-Primer: the construction rules (purpose, scope, lenses, validation criteria).<\/li>\n\n\n\n
                                            • Topical Primers: domain-specific standards, evidence gates, thresholds, escalation triggers, and audit requirements.<\/li>\n<\/ul>\n\n\n\n

                                              3) Execution Layer \u2014 Romer<\/strong><\/p>\n\n\n\n

                                                \n
                                              • Translates the Primer into executable reasoning: choose lens (R_logic, R_prob, R_hybrid, etc.), run decision nodes, apply weights\/thresholds, and produce a traceable<\/em> result.<\/li>\n\n\n\n
                                              • Explicit human-in-the-loop triggers (confidence < floor, boundary cases, bias flags).<\/li>\n<\/ul>\n\n\n\n

                                                4) Orchestration Layer \u2014 Process<\/strong><\/p>\n\n\n\n

                                                  \n
                                                • Intake \u2192 Primer selection \u2192 Romer activation \u2192 Validation gates \u2192 Conditional human review \u2192 Finalization \u2192 Learning capture.<\/li>\n\n\n\n
                                                • Ensures repeatability, variance reduction, and governance in practice.<\/li>\n<\/ul>\n\n\n\n

                                                  5) Assurance & Trace Layer<\/strong><\/p>\n\n\n\n

                                                    \n
                                                  • Romer Trace provides the black-box recorder for reasoning; audit mode surfaces full chains with overrides and rationale.<\/li>\n\n\n\n
                                                  • Learning capture converts edge cases into bounded updates.<\/li>\n<\/ul>\n\n\n\n

                                                    IFAAMAS-Compatible by Design<\/h3>\n\n\n\n

                                                    XPlain embeds AAMAS concepts without locking you to a specific runtime:<\/p>\n\n\n\n

                                                      \n
                                                    • BDI Alignment:<\/strong>\n
                                                        \n
                                                      • Beliefs<\/em> \u2194 evidence + lenses in Primers<\/li>\n\n\n\n
                                                      • Desires<\/em> \u2194 stated objectives and metrics<\/li>\n\n\n\n
                                                      • Intentions<\/em> \u2194 Romer plan selection + commitment rules, logged in the trace
                                                        This makes agent goals and commitments explicit and verifiable.<\/li>\n<\/ul>\n<\/li>\n\n\n\n
                                                      • Norms & Governance:<\/strong>
                                                        Evidence gates, pause points, and change budgets operationalize institutional norms and oversight\u2014core AAMAS themes adapted to modern LLM agents.<\/li>\n\n\n\n
                                                      • Protocols & Coordination:<\/strong>
                                                        Process handoffs and audit-ready traces provide the substrate for formal interaction patterns (e.g., you can layer FIPA\/contract-net semantics over the same gated decisions).<\/li>\n<\/ul>\n\n\n\n
                                                        \n

                                                        Bottom line: XPlain is the integration surface<\/em>\u2014bring BDI, norms, or mechanism-design ideas from AAMAS, and bind them to evidence, gates, and traces for real-world assurance.<\/p>\n<\/blockquote>\n\n\n\n

                                                        \n\n\n\n

                                                        What You Get (Out-of-the-Box Behaviors)<\/h3>\n\n\n\n
                                                          \n
                                                        • Consistency & comparability:<\/strong> identical inputs \u2192 identical governed logic.<\/li>\n\n\n\n
                                                        • Human-in-the-loop by rule, not by hope:<\/strong> explicit triggers for review and escalation.<\/li>\n\n\n\n
                                                        • Dual-reader outputs:<\/strong> concise business report + detailed system report + full trace on demand.<\/li>\n<\/ul>\n\n\n\n
                                                          \n\n\n\n

                                                          Example (Abbreviated)<\/h3>\n\n\n\n
                                                            \n
                                                          • Primer:<\/strong> Third-Party Security (v2.1) \u2014 mandatory evidence, confidence floor, and constraints to reference ISO\/NIST.<\/li>\n\n\n\n
                                                          • Romer:<\/strong> R_hybrid lens applies scoring and thresholds; boundary cases escalate.<\/li>\n\n\n\n
                                                          • Output:<\/strong> Risk score with confidence, evidence summary, and audit trail pointer\u2014repeatable across vendors.<\/li>\n<\/ul>\n\n\n\n
                                                            \n\n\n\n

                                                            Governance Note<\/h3>\n\n\n\n

                                                            This page reflects XPlain\u2019s interpretive architecture and permitted artifacts (Primers, Romers, Traces, Processes). Executable analytics or data-run simulations fall outside the governed scope unless explicitly authorized.<\/p>\n\n\n\n

                                                            CTA<\/h3>\n\n\n\n

                                                            Design your first Primer<\/strong> \u2192 Encode your domain rules, evidence gates, and assurance thresholds, then derive a Romer to make your agent auditable from day one.<\/p>\n\n\n\n

                                                            <\/p>","protected":false},"excerpt":{"rendered":"

                                                            XPlain is a general-purpose form of agentic AI that\u2019s platform-agnostic and audit-ready. It orchestrates decisions using Meta-Primers, domain Primers, and Romers to align beliefs, goals, and commitments with evidence gates, human pause points, and assurance controls. The result: agents you can trust\u2014traceable, compliant, and ready for real-world deployment across GRC, supply chain, risk, and data […]<\/p>\n","protected":false},"author":2,"featured_media":0,"parent":132,"menu_order":1,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-304","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/pages\/304","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/comments?post=304"}],"version-history":[{"count":17,"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/pages\/304\/revisions"}],"predecessor-version":[{"id":468,"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/pages\/304\/revisions\/468"}],"up":[{"embeddable":true,"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/pages\/132"}],"wp:attachment":[{"href":"https:\/\/xplain-r.com\/fr\/wp-json\/wp\/v2\/media?parent=304"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}