Explain AI's end-to-end banking QE contribution with synchronized narration

README.md

@@ -7,10 +7,10 @@ that collects evidence and reusable method for AI-assisted quality engineering i
 
 The homepage includes two audience presentations:
 
-- [Executive strategic vision](https://tomqwu.github.io/ai_qe/briefings/evp/): 25 slides on
+- [Executive strategic vision](https://tomqwu.github.io/ai_qe/briefings/evp/): 26 slides on
   industry outlook, strategic vision, value, ownership and capability expansion.
 - [Technical delivery architecture](https://tomqwu.github.io/ai_qe/briefings/technical/):
-  35 slides on context, test validity, evaluation, agent controls, deployment, recovery and integration.
+  36 slides on context, test validity, evaluation, agent controls, deployment, recovery and integration.
 
 These are research-informed perspectives and a proposed architecture, not measured bank results.
 
@@ -60,7 +60,7 @@ the site edition; source review dates remain attached to their individual resear
 
 ## Maintaining the briefings
 
-The [fintech case explorer](https://tomqwu.github.io/ai_qe/case-studies/fintech/) has its own 19-slide executive and 30-slide technical decks. `_data/fintech_case.json` holds the fictional assumptions, workflow, sources and maturity profiles. `_data/fintech_decks.json` holds the slide narratives; `_includes/fintech/` contains their editable diagrams and shared layouts. Keep this illustrative case distinct from observed industry findings.
+The [fintech case explorer](https://tomqwu.github.io/ai_qe/case-studies/fintech/) has its own 21-slide executive and 33-slide technical decks. `_data/fintech_case.json` holds the fictional assumptions, workflow, sources and maturity profiles. `_data/fintech_decks.json` holds the slide narratives; `_includes/fintech/` contains their editable diagrams and shared layouts. Keep this illustrative case distinct from observed industry findings.
 
 Run `node tools/export_decks.cjs --fintech` against the preview to export the case PDFs. The edition in the case data, deck front matter and `fintech_edition` release field must agree. Run `node tools/fintech-browser-test.cjs` and `python tools/verify_fintech.py _site` after the build. `npm test` includes capacity and payment-model checks. The browser case is a deterministic teaching model; it does not connect to a payment service or invoke the example QA frameworks.
 
@@ -233,3 +233,5 @@ Fintech decks accept `?route=client#slide-N` for a curated sequence. Original an
 ## AI adoption roadmap
 
 `_data/ai_adoption.json` defines four capability layers, PAY-142 examples, authority boundaries, prerequisites, evidence gates and reviewed primary sources. `/ai-adoption/` provides an interactive leadership view; the same native diagrams appear on fintech executive slide 17 and technical slide 28. Layer 4 supports reuse of layers 1–3; shared foundations start at every layer and agent authority is optional. This is an authored roadmap, not a client maturity assessment. Run `node tools/ai-adoption-browser-test.cjs` for selection, history, readiness links, responsive layout, search and no-JavaScript access.
+
+The banking AI walkthrough uses `_data/qe_journey.json` for nine explicit handoffs and the three-state comparison. Caption-anchored profiles in `assets/data/narration-flows.json` synchronize the same visuals on pages and slides. The QA effort model retains its eight stages; developer unit work has no added savings assumption.

_data/briefing_room.json

@@ -9,7 +9,7 @@
     "url": "/briefings/fintech-evp/",
     "pdf_prefix": "ai-qe-fintech-evp",
     "edition": "fintech_edition",
-    "slides": 19,
+    "slides": 21,
     "time": "25–30 min",
     "outline": [
       "The payment story and manual-work baseline",
@@ -22,7 +22,8 @@
       "Before / after: manual API work and the target pilot workflow",
       "Observed time framework for the API pilot and brainstorm questions",
       "Four AI adoption layers: individual copilots, connected workflows, bounded agents and shared platform services",
-      "Before/after architecture and outcomes beyond hours · slides 18–19"
+      "Before/after architecture and outcomes beyond hours · slides 18–19",
+      "AI contribution and artifact handoffs · slides 20–21"
     ]
   },
   {
@@ -35,15 +36,16 @@
     "url": "/briefings/fintech-technical/",
     "pdf_prefix": "ai-qe-fintech-technical",
     "edition": "fintech_edition",
-    "slides": 30,
+    "slides": 33,
     "time": "35–45 min",
     "outline": [
       "System boundaries and payment API contract · slides 1–5",
       "Versioned context, fixtures, assertions and CI artifacts · slides 6–12",
       "Failure injection, readiness proof and promotion rules · slides 13–19",
       "Virtualization, deployment and candidate validation · slides 20–25",
       "Adapter contracts, before/after and AI integration layers · slides 26–28",
-      "Before/after architecture and outcomes beyond hours · slides 27, 29–30"
+      "Before/after architecture and outcomes beyond hours · slides 27, 29–30",
+      "Nine-step workflow, artifact trace and three-state comparison · slides 31–33"
     ]
   },
   {
@@ -56,15 +58,16 @@
     "url": "/briefings/evp/",
     "pdf_prefix": "ai-qe-evp",
     "edition": "slide_edition",
-    "slides": 25,
+    "slides": 26,
     "time": "30–40 min",
     "outline": [
       "Vision and value · slides 1–5",
       "Industry, capability and ownership · slides 6–14",
       "Roadmap, leadership and proof · slides 15–21",
       "Client adoption assumptions and foundation ownership",
       "QE modernization, repeatable environments and virtual services",
-      "Financial-services evidence and a focused client trial"
+      "Financial-services evidence and a focused client trial",
+      "AI-assisted banking QE: end-to-end workflow · final slide"
     ]
   },
   {
@@ -77,15 +80,16 @@
     "url": "/briefings/technical/",
     "pdf_prefix": "ai-qe-technical",
     "edition": "slide_edition",
-    "slides": 35,
+    "slides": 36,
     "time": "45–60 min",
     "outline": [
       "Architecture, context and test validity · slides 1–12",
       "Evaluation, controls and operations · slides 13–21",
       "Integration, delivery and failure resolution · slides 22–29",
       "Client adoption assumptions and foundation ownership",
       "QE modernization, repeatable environments and virtual services",
-      "Financial-services evidence and a focused client trial"
+      "Financial-services evidence and a focused client trial",
+      "AI-assisted banking QE: end-to-end workflow · final slide"
     ]
   }
 ]

_data/fintech_case.json

@@ -1,5 +1,5 @@
 {
-  "edition": "1.16.1",
+  "edition": "1.22.0",
   "reviewed": "2026-09-07",
   "adoptionAssumptions": "/platform-readiness/?preset=harbor",
   "company": "Our Banking Client",

_data/fintech_decks.json

@@ -56,14 +56,14 @@
       "chapter": "The proposed change"
     },
     {
-      "title": "Before / after: the API testing workflow",
+      "title": "Before / after: separate modernization from AI",
       "visual": "before-after",
       "lead": "",
       "rows": null,
       "columns": null,
       "points": null,
-      "note": "Pilot target; validate foundations and measure effort, review, rework and quality before claiming gains.",
-      "sources": "S1,S2,S5,S6,S7",
+      "note": "Same payment scope. The two target states are proposed; no productivity or quality improvement has been measured.",
+      "sources": "AL1,AL2,S5,S6,S7",
       "chapter": ""
     },
     {
@@ -253,6 +253,28 @@
       "note": "Business aims: fewer escaped payment defects and emergency fixes. These proposed pilot measures do not establish those outcomes.",
       "sources": "QO1,QO2",
       "chapter": "Architecture and outcomes"
+    },
+    {
+      "title": "What AI contributes, from requirement to release",
+      "visual": "ai-journey",
+      "lead": "",
+      "rows": null,
+      "columns": null,
+      "points": null,
+      "note": "PAY-142 is an authored scenario. Each AI output needs review; test runners execute and people decide.",
+      "sources": "AL1,AL2,S5,S6,S7",
+      "chapter": "AI through the release"
+    },
+    {
+      "title": "Follow one payment test through the handoffs",
+      "visual": "artifact-trace",
+      "lead": "",
+      "rows": null,
+      "columns": null,
+      "points": null,
+      "note": "Illustrative artifacts and outcomes. Preserve failed evidence; a passing retest alone does not approve release.",
+      "sources": "AL1,AL2,S5,S6,S7",
+      "chapter": "AI through the release"
     }
   ],
   "technical": [
@@ -360,6 +382,10 @@
       "visual": null,
       "lead": "",
       "rows": [
+        [
+          "Developer unit",
+          "Copilot drafts JUnit tests for idempotency logic; developers review. A fake repository cannot prove real payment integration."
+        ],
         [
           "Service and API",
           "REST Assured + JUnit. Cover retry, concurrent requests, idempotency conflicts and journal outcomes."
@@ -383,7 +409,7 @@
       ],
       "points": null,
       "note": "Test distribution follows behavior and observability. A new browser framework does not solve a shared settlement environment.",
-      "sources": "S6,S8",
+      "sources": "AL1,S6,S8",
       "chapter": ""
     },
     {
@@ -734,6 +760,39 @@
       "sources": "QO2",
       "chapter": "Outcomes beyond hours",
       "metric_role": "supporting"
+    },
+    {
+      "title": "Nine AI-assisted steps, with explicit handoffs",
+      "visual": "ai-journey",
+      "lead": "",
+      "rows": null,
+      "columns": null,
+      "points": null,
+      "note": "AI drafts; reviewers approve; CI executes. Context, artifacts and evidence remain versioned.",
+      "sources": "AL1,AL2,S5,S6,S7",
+      "chapter": "AI through the release"
+    },
+    {
+      "title": "PAY-142: requirement, test, failure, fix and proof",
+      "visual": "artifact-trace",
+      "lead": "",
+      "rows": null,
+      "columns": null,
+      "points": null,
+      "note": "Authored trace IDs. Fixed assertions: original transfer ID, one journal, payer 90000 and recipient 10000 minor units.",
+      "sources": "AL1,AL2,S5,S6,S7",
+      "chapter": "AI through the release"
+    },
+    {
+      "title": "Three states: isolate the incremental AI contribution",
+      "visual": "before-after",
+      "lead": "",
+      "rows": null,
+      "columns": null,
+      "points": null,
+      "note": "Compare equivalent payment packs. Retain real-provider checks and measure review, rework and platform overhead.",
+      "sources": "AL1,AL2,S5,S6,S7",
+      "chapter": "AI through the release"
     }
   ]
 }