Fix illustration format: use standard class and figure numbering

Change class="opener-illustration" to class="illustration" (matches
CSS rules in book.css) and add <strong>Figure X.Y.Z</strong> numbering
to all 13 figcaptions in Part XI sections.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: apartsin

part-11-idea-to-product/module-36-idea-to-product/section-36.1.html

@@ -27,11 +27,11 @@ <h1>What Makes AI Products Different</h1>
         <cite><span class="agent-avatar-inline" style="background-color: #2c3e50;"><img src="../../front-matter/images/agents/dropout.png" alt="Deploy"></span> <a href="../../front-matter/wisdom-council.html#deploy">Deploy</a> <span class="agent-desc">The Production Realist</span></cite>
     </blockquote>
 
-    <figure class="opener-illustration">
+    <figure class="illustration">
         <img src="images/ai-product-differences.png"
              alt="A traditional vending machine next to a glowing AI oracle booth with probability clouds, with a developer standing between them holding a product roadmap."
              loading="lazy">
-        <figcaption>Traditional software delivers deterministic outputs; AI products live in a world of probabilities, confidence scores, and graceful degradation.</figcaption>
+        <figcaption><strong>Figure 36.1.1</strong>: Traditional software delivers deterministic outputs; AI products live in a world of probabilities, confidence scores, and graceful degradation.</figcaption>
     </figure>
 
     <div class="prerequisites">

part-11-idea-to-product/module-36-idea-to-product/section-36.2.html

@@ -27,11 +27,11 @@ <h1>Choosing the Model's Role</h1>
         <cite><span class="agent-avatar-inline" style="background-color: #2c3e50;"><img src="../../front-matter/images/agents/dropout.png" alt="Deploy"></span> <a href="../../front-matter/wisdom-council.html#deploy">Deploy</a> <span class="agent-desc">The Production Realist</span></cite>
     </blockquote>
 
-    <figure class="opener-illustration">
+    <figure class="illustration">
         <img src="images/model-role-spectrum.png"
              alt="A horizontal spectrum dial from Copilot to Autopilot, with five role icons below: pencil for drafter, tag for classifier, signpost for router, magnifying glass for researcher, checkmark shield for verifier."
              loading="lazy">
-        <figcaption>Every AI product sits somewhere on the copilot-to-autopilot spectrum. The role you assign determines the architecture, the risk profile, and the trust your users need to place in the system.</figcaption>
+        <figcaption><strong>Figure 36.2.1</strong>: Every AI product sits somewhere on the copilot-to-autopilot spectrum. The role you assign determines the architecture, the risk profile, and the trust your users need to place in the system.</figcaption>
     </figure>
 
     <div class="prerequisites">

part-11-idea-to-product/module-36-idea-to-product/section-36.3.html

@@ -27,11 +27,11 @@ <h1>Risk and Feasibility Assessment</h1>
         <cite><span class="agent-avatar-inline" style="background-color: #2c3e50;"><img src="../../front-matter/images/agents/dropout.png" alt="Deploy"></span> <a href="../../front-matter/wisdom-council.html#deploy">Deploy</a> <span class="agent-desc">The Production Realist</span></cite>
     </blockquote>
 
-    <figure class="opener-illustration">
+    <figure class="illustration">
         <img src="images/risk-feasibility.png"
              alt="A traffic light with risk, assessment, and feasibility icons in its three lights, with a product manager holding a scorecard beside it."
              loading="lazy">
-        <figcaption>Before committing engineering resources, score every AI product idea across technical, data, regulatory, and economic dimensions. The Feasibility Scorecard makes hidden risks visible.</figcaption>
+        <figcaption><strong>Figure 36.3.1</strong>: Before committing engineering resources, score every AI product idea across technical, data, regulatory, and economic dimensions. The Feasibility Scorecard makes hidden risks visible.</figcaption>
     </figure>
 
     <div class="prerequisites">

part-11-idea-to-product/module-36-idea-to-product/section-36.4.html

@@ -27,11 +27,11 @@ <h1>Case Studies: Role Assignment in Practice</h1>
         <cite><span class="agent-avatar-inline" style="background-color: #2c3e50;"><img src="../../front-matter/images/agents/dropout.png" alt="Deploy"></span> <a href="../../front-matter/wisdom-council.html#deploy">Deploy</a> <span class="agent-desc">The Production Realist</span></cite>
     </blockquote>
 
-    <figure class="opener-illustration">
+    <figure class="illustration">
         <img src="images/case-studies-roles.png"
              alt="Four open case files on a desk showing different industry scenarios: hospital triage, e-commerce ranking, legal review, and customer support routing, with a magnifying glass hovering above."
              loading="lazy">
-        <figcaption>Four real-world case studies show how role assignment, feasibility scoring, and iterative refinement play out across healthcare, e-commerce, legal, and customer support domains.</figcaption>
+        <figcaption><strong>Figure 36.4.1</strong>: Four real-world case studies show how role assignment, feasibility scoring, and iterative refinement play out across healthcare, e-commerce, legal, and customer support domains.</figcaption>
     </figure>
 
     <div class="prerequisites">

part-11-idea-to-product/module-37-building-steering/section-37.1.html

@@ -27,11 +27,11 @@ <h1>The Observe-Steer Development Loop</h1>
         <cite>Andrej Karpathy, on "vibe coding" (2025)</cite>
     </blockquote>
 
-    <figure class="opener-illustration">
+    <figure class="illustration">
         <img src="images/vibe-coding-loop.png"
              alt="A developer at a glowing monitor surrounded by a circular loop of colorful arrows: observe, steer, generate, document."
              loading="lazy">
-        <figcaption>The observe-steer loop replaces the traditional spec-build-test cycle. You describe intent, observe the AI's output, and steer iteratively toward the goal.</figcaption>
+        <figcaption><strong>Figure 37.1.1</strong>: The observe-steer loop replaces the traditional spec-build-test cycle. You describe intent, observe the AI's output, and steer iteratively toward the goal.</figcaption>
     </figure>
 
     <div class="prerequisites">

part-11-idea-to-product/module-37-building-steering/section-37.2.html

@@ -27,11 +27,11 @@ <h1>The Founder's Prototype Loop</h1>
         <cite><span class="agent-avatar-inline" style="background-color: #2c3e50;"><img src="../../front-matter/images/agents/dropout.png" alt="Deploy"></span> <a href="../../front-matter/wisdom-council.html#deploy">Deploy</a> <span class="agent-desc">The Production Realist</span></cite>
     </blockquote>
 
-    <figure class="opener-illustration">
+    <figure class="illustration">
         <img src="images/prototype-loop.png"
              alt="A vertical rocket launch pad built from stacked colorful building blocks, with a founder at the base and circular arrows labeled Build, Measure, Steer."
              loading="lazy">
-        <figcaption>The founder's prototype loop builds evidence at every iteration: a vertical-slice demo that exercises the full stack, measured against explicit quality gates.</figcaption>
+        <figcaption><strong>Figure 37.2.1</strong>: The founder's prototype loop builds evidence at every iteration: a vertical-slice demo that exercises the full stack, measured against explicit quality gates.</figcaption>
     </figure>
 
     <div class="prerequisites">

part-11-idea-to-product/module-37-building-steering/section-37.3.html

@@ -27,11 +27,11 @@ <h1>Documentation as Control Surface</h1>
         <cite>Adapted from industry practice, 2025</cite>
     </blockquote>
 
-    <figure class="opener-illustration">
+    <figure class="illustration">
         <img src="images/documentation-control.png"
              alt="A large open book functioning as a control panel with knobs, switches, and indicator lights instead of text, with a developer and an AI robot on opposite sides."
              loading="lazy">
-        <figcaption>When AI writes the code, documentation shifts from explanation to control: intent capture, decision tracking, and machine-readable specifications that guide future AI actions.</figcaption>
+        <figcaption><strong>Figure 37.3.1</strong>: When AI writes the code, documentation shifts from explanation to control: intent capture, decision tracking, and machine-readable specifications that guide future AI actions.</figcaption>
     </figure>
 
     <div class="prerequisites">

part-11-idea-to-product/module-37-building-steering/section-37.4.html

@@ -27,11 +27,11 @@ <h1>AI Coding Assistants: Trust but Verify</h1>
         <cite>Adapted from "The Strange Future of AI-Written Software" (2025)</cite>
     </blockquote>
 
-    <figure class="opener-illustration">
+    <figure class="illustration">
         <img src="images/ai-coding-trust-verify.png"
              alt="A human developer and a small AI robot pair-programming at a shared desk, with the human holding a magnifying glass and a trust meter between them."
              loading="lazy">
-        <figcaption>AI coding assistants shift the developer's role from writer to evaluator. The trust-but-verify discipline keeps velocity high while catching the errors that AI introduces.</figcaption>
+        <figcaption><strong>Figure 37.4.1</strong>: AI coding assistants shift the developer's role from writer to evaluator. The trust-but-verify discipline keeps velocity high while catching the errors that AI introduces.</figcaption>
     </figure>
 
     <div class="prerequisites">

part-11-idea-to-product/module-37-building-steering/section-37.5.html

@@ -27,11 +27,11 @@ <h1>From Prototype to MVP</h1>
         <cite>Adapted from Eric Ries, The Lean Startup (2011)</cite>
     </blockquote>
 
-    <figure class="opener-illustration">
+    <figure class="illustration">
         <img src="images/prototype-to-mvp.png"
              alt="A bridge connecting two floating islands: a rough prototype workshop on the left and a polished product factory on the right, with quality gate checkpoints along the bridge."
              loading="lazy">
-        <figcaption>The prototype-to-MVP bridge has quality gates at every stage: evaluation contracts, monitoring readiness, fallback strategies, and user feedback loops must all be in place before crossing.</figcaption>
+        <figcaption><strong>Figure 37.5.1</strong>: The prototype-to-MVP bridge has quality gates at every stage: evaluation contracts, monitoring readiness, fallback strategies, and user feedback loops must all be in place before crossing.</figcaption>
     </figure>
 
     <div class="prerequisites">

part-11-idea-to-product/module-38-shipping-scaling/section-38.1.html

@@ -27,11 +27,11 @@ <h1>Launch Constraints and AI Unit Economics</h1>
         <cite><span class="agent-avatar-inline" style="background-color: #2c3e50;"><img src="../../front-matter/images/agents/dropout.png" alt="Deploy"></span> <a href="../../front-matter/wisdom-council.html#deploy">Deploy</a> <span class="agent-desc">The Production Realist</span></cite>
     </blockquote>
 
-    <figure class="opener-illustration">
+    <figure class="illustration">
         <img src="images/launch-economics.png"
              alt="A balance scale with a glowing AI brain on one side and a stack of gold coins with token symbols on the other, with a startup founder peering through a magnifying glass."
              loading="lazy">
-        <figcaption>Launch economics for AI products revolve around the balance between capability and cost. Token billing, platform choices, and compliance readiness determine whether your product is viable at scale.</figcaption>
+        <figcaption><strong>Figure 38.1.1</strong>: Launch economics for AI products revolve around the balance between capability and cost. Token billing, platform choices, and compliance readiness determine whether your product is viable at scale.</figcaption>
     </figure>
 
     <div class="prerequisites">