cluesmith · waleedkadous · Feb 24, 2026 · Feb 23, 2026 · Feb 23, 2026 · Feb 23, 2026
diff --git a/marketing/differentiators-vs-claude-code.md b/marketing/differentiators-vs-claude-code.md
@@ -0,0 +1,13 @@
+# Codev vs Claude Code: Key Differentiators
+
+1. **Multi-model**: We use Gemini + Codex as reviewers alongside Claude. Each catches different classes of issues — Codex finds security edge cases, Claude catches runtime semantics, Gemini catches architecture problems.
+
+2. **Specs and plans are first-class citizens**: Every feature produces a specification and an implementation plan that are preserved as project artifacts. You always know WHY something was built and HOW it was designed.
+
+3. **Plans are enforced**: You can't start Phase 2 until the acceptance criteria for Phase 1 are met. The Porch state machine makes the process deterministic — no skipping steps. Human gates require architect approval before implementation begins, so the AI can't run off and build the wrong thing. Testing is enforced as acceptance criteria, not optional — this is why codev produces 2.9x more test coverage consistently.
+
+4. **Annotation over direct editing**: It's far more about annotating docs than directly editing code. Reviews, specs, and plans are documents that guide the work rather than the AI just hacking at files.
+
+5. **Parallel builders**: Push parallelization to its limit by having one "Architect" AI that can direct 5+ builders working simultaneously in isolated worktrees.
+
+6. **Whole lifecycle management**: From idea through specification, planning, implementation, review, PR, and deployment — codev manages the entire lifecycle, not just the coding step.
diff --git a/marketing/logo-transparent.png b/marketing/logo-transparent.png
diff --git a/marketing/logo.png b/marketing/logo.png
diff --git a/marketing/real-results.md b/marketing/real-results.md
@@ -0,0 +1,24 @@
+# Real Results
+
+## Head-to-Head: Codev vs Claude Code (R4 Comparison)
+
+- **Overall quality score**: 7.0 vs 5.8
+- **Test coverage**: 2.9x more test lines (0.79:1 vs 0.26:1 test-to-code ratio)
+- **Fewer bugs**: Fewer consensus bugs, fewer High severity issues
+- **Deployment readiness**: Codev produces Dockerfile, .dockerignore, deploy README. Claude Code produces none.
+- **Architecture**: Clean three-layer separation vs single-flow mixing parsing and execution
+
+## 14-Day Production Sprint
+
+- **106 PRs merged** in 14 days (53/week)
+- **85% fully autonomous** — builders completed without human intervention
+- **24 pre-merge bug catches**, 4 security-critical
+- **$1.59 per PR** ($168.64 total), 3.4x ROI
+- **66% of bugfixes** ship in under 30 minutes
+
+## Multi-Model Review Catches What Single-Model Misses
+
+- **Codex**: Security edge cases
+- **Claude**: Runtime semantics
+- **Gemini**: Architecture problems
+- Three independent reviewers with complementary blind spots
diff --git a/marketing/slides-intro.html b/marketing/slides-intro.html
@@ -0,0 +1,246 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+<meta charset="UTF-8">
+<meta name="viewport" content="width=device-width, initial-scale=1.0">
+<title>codev — Demo Intro</title>
+<style>
+  @import url('https://fonts.googleapis.com/css2?family=Source+Sans+3:wght@300;400;600;700;900&display=swap');
+
+  * { margin: 0; padding: 0; box-sizing: border-box; }
+
+  body {
+    font-family: 'Source Sans 3', sans-serif;
+    background: #1a1a1a;
+    color: #e4e4e7;
+    overflow: hidden;
+  }
+
+  .slide {
+    display: none;
+    width: 100vw;
+    height: 100vh;
+    padding: 80px 100px;
+    flex-direction: column;
+    justify-content: center;
+  }
+
+  .slide.active { display: flex; }
+
+  /* ─── Title logo (large, on title slide only) ─── */
+  .logo-title {
+    display: flex;
+    align-items: center;
+    gap: 24px;
+    margin-bottom: 64px;
+  }
+  .logo-title img { width: 88px; height: 88px; }
+  .logo-title span {
+    font-size: 52px;
+    font-weight: 300;
+    letter-spacing: 0.08em;
+    color: #e4e4e7;
+  }
+
+  /* ─── Footer logo (every slide, bottom-right) ─── */
+  .footer-logo {
+    position: fixed;
+    bottom: 32px;
+    right: 40px;
+    display: flex;
+    align-items: center;
+    gap: 10px;
+  }
+  .footer-logo img { width: 32px; height: 32px; opacity: 0.5; }
+  .footer-logo span {
+    font-size: 20px;
+    font-weight: 300;
+    letter-spacing: 0.08em;
+    color: #52525b;
+  }
+
+  /* ─── Typography ─── */
+  h1 {
+    font-size: 80px;
+    font-weight: 900;
+    line-height: 1.05;
+    margin-bottom: 40px;
+    color: #fafafa;
+  }
+  h1 .accent { color: #818cf8; }
+
+  h2 {
+    font-size: 64px;
+    font-weight: 700;
+    line-height: 1.1;
+    margin-bottom: 48px;
+    color: #fafafa;
+  }
+  h2 .accent { color: #818cf8; }
+
+  .subtitle {
+    font-size: 36px;
+    font-weight: 300;
+    color: #a1a1aa;
+    line-height: 1.5;
+    max-width: 960px;
+  }
+
+  .byline {
+    font-size: 28px;
+    font-weight: 400;
+    color: #71717a;
+    margin-top: 64px;
+  }
+  .byline strong { color: #e4e4e7; font-weight: 600; }
+
+  /* ─── Bullet lists ─── */
+  .bullets {
+    list-style: none;
+    display: flex;
+    flex-direction: column;
+    gap: 40px;
+  }
+  .bullets li {
+    font-size: 36px;
+    font-weight: 400;
+    line-height: 1.4;
+    color: #d4d4d8;
+    padding-left: 40px;
+    position: relative;
+  }
+  .bullets li::before {
+    content: '';
+    position: absolute;
+    left: 0;
+    top: 16px;
+    width: 14px;
+    height: 14px;
+    border-radius: 50%;
+    background: #818cf8;
+  }
+  .bullets li strong { color: #fafafa; }
+
+  /* ─── Stat callout ─── */
+  .stat {
+    font-size: 80px;
+    font-weight: 700;
+    color: #818cf8;
+    line-height: 1;
+    margin-bottom: 10px;
+  }
+  .stat-label {
+    font-size: 26px;
+    color: #71717a;
+    font-weight: 400;
+  }
+
+  .stats-grid {
+    display: grid;
+    grid-template-columns: 1fr 1fr 1fr 1fr;
+    gap: 48px;
+    margin-top: 56px;
+  }
+
+  /* ─── Navigation ─── */
+  .slide-counter {
+    position: fixed;
+    bottom: 32px;
+    left: 40px;
+    font-size: 14px;
+    color: #333;
+    letter-spacing: 0.04em;
+  }
+</style>
+</head>
+<body>
+
+<!-- ═══════════════════════════════════════════
+     SLIDE 1: Title
+     ═══════════════════════════════════════════ -->
+<div class="slide active" id="slide-1">
+  <div class="logo-title">
+    <img src="logo-transparent.png" alt="codev">
+    <span>codev</span>
+  </div>
+
+  <h1>Productive Human-AI Co-development<br>on Large Codebases: <span class="accent">Demo</span></h1>
+
+  <p class="byline"><strong>Waleed Kadous</strong></p>
+</div>
+
+<!-- ═══════════════════════════════════════════
+     SLIDE 2: The Problem
+     ═══════════════════════════════════════════ -->
+<div class="slide" id="slide-2">
+  <h2>Vibe coding is not <span class="accent">production</span> coding</h2>
+
+  <ul class="bullets">
+    <li>We've all seen the demos — "build me an app" in minutes</li>
+    <li><strong>Production is different</strong> — tests, review, deployment, docs, maintenance</li>
+    <li>Real codebases are 10K-100K+ lines. Context windows run out.</li>
+    <li><strong>Codev</strong> is a Human-AI development OS built for this reality</li>
+  </ul>
+</div>
+
+<!-- ═══════════════════════════════════════════
+     SLIDE 3: Results
+     ═══════════════════════════════════════════ -->
+<div class="slide" id="slide-3">
+  <h2>We used Codev to build <span class="accent">Codev</span></h2>
+
+  <p class="subtitle" style="margin-bottom: 40px;">~80K lines of TypeScript. 289 source files. 14 days.</p>
+
+  <ul class="bullets">
+    <li><strong>106 PRs merged in 14 days</strong> — 85% fully autonomous</li>
+    <li><strong>20 bugs caught before merge</strong> — including 1 security-critical</li>
+    <li><strong>26 features shipped</strong> — from session management to workspace orchestration</li>
+    <li>Throughput of <strong>3-4 elite engineers</strong> (vs LinearB 2026 benchmarks, 8.1M PRs)</li>
+    <li>With everything you'd expect: <strong>testing, PRs, architecture overview, design docs</strong></li>
+  </ul>
+</div>
+
+<!-- ═══════════════════════════════════════════
+     SLIDE 4: Let me show you
+     ═══════════════════════════════════════════ -->
+<div class="slide" id="slide-4">
+  <div class="logo-title">
+    <img src="logo-transparent.png" alt="codev">
+    <span>codev</span>
+  </div>
+
+  <h1>Let me show you</h1>
+</div>
+
+<!-- ═══════════════════════════════════════════ -->
+<div class="footer-logo">
+  <img src="logo-transparent.png" alt="codev">
+  <span>codev</span>
+</div>
+<div class="slide-counter">1 / 4</div>
+
+<script>
+  const slides = document.querySelectorAll('.slide');
+  const counter = document.querySelector('.slide-counter');
+  let current = 0;
+
+  function show(i) {
+    slides[current].classList.remove('active');
+    current = Math.max(0, Math.min(i, slides.length - 1));
+    slides[current].classList.add('active');
+    counter.textContent = `${current + 1} / ${slides.length}`;
+  }
+
+  document.addEventListener('keydown', e => {
+    if (e.key === 'ArrowRight' || e.key === ' ') show(current + 1);
+    if (e.key === 'ArrowLeft') show(current - 1);
+  });
+
+  document.addEventListener('click', e => {
+    if (e.clientX > window.innerWidth / 2) show(current + 1);
+    else show(current - 1);
+  });
+</script>
+
+</body>
+</html>
diff --git a/marketing/why-pay-attention.md b/marketing/why-pay-attention.md
@@ -0,0 +1,31 @@
+# Why You Should Pay Attention
+
+## When Starting a New Project
+
+We ran the same feature spec through Claude Code and Codev four times. Codev produces far fewer bugs, higher quality code, and manages the whole lifecycle — tests, deployment artifacts, and PRs.
+
+| Dimension | Claude Code | Codev | Delta |
+|-----------|:----------:|:-----:|:-----:|
+| **Bugs** | 6.7 | 7.3 | +0.7 |
+| **Code Quality** | 7.0 | 7.7 | +0.7 |
+| **Maintainability** | 7.3 | 7.3 | 0.0 |
+| **Tests** | 5.0 | 6.7 | +1.7 |
+| **Extensibility** | 5.7 | 6.3 | +0.7 |
+| **NL Interface** | 6.3 | 6.7 | +0.3 |
+| **Deployment** | 2.7 | 6.7 | +4.0 |
+| **Overall** | **5.8** | **7.0** | **+1.2** |
+
+Scored by three independent AI reviewers (Claude, Codex, Gemini). Full methodology in the R4 comparison report.
+
+**No free lunch**: Codev took ~56 minutes vs ~15 minutes for Claude Code, and cost 3-5x more ($14-19 vs $4-7) due to multi-model review overhead. The question is whether the quality delta is worth it for your project.
+
+## When Maintaining Large Codebases
+
+We used Codev to build Codev — an ~80K-line TypeScript codebase across 289 source files.
+
+- **106 PRs merged in 14 days** — 85% completed fully autonomously, no human intervention
+- **20 bugs caught before merge** by multi-model review, including 1 security-critical
+- **Extensive architectural documentation**, including an accessible `arch.md` that stays current as the codebase evolves
+- **Allowed us to ship 26 features in two weeks** — from custom session management to full workspace orchestration
+- **Equivalent throughput of 3-4 elite engineers** at $1.59 per PR ($168.64 total for the sprint)
+- But with everything you'd expect of production grade: testing, PRs, clear docs, and multi-model code review on every change
diff --git a/packages/codev/dashboard/__tests__/analytics.test.tsx b/packages/codev/dashboard/__tests__/analytics.test.tsx
@@ -27,10 +27,9 @@ function makeStats(overrides: Partial<AnalyticsResponse> = {}): AnalyticsRespons
     timeRange: '7d',
     activity: {
       prsMerged: 12,
-      avgTimeToMergeHours: 3.5,
+      medianTimeToMergeHours: 3.5,
       issuesClosed: 6,
-      avgTimeToCloseBugsHours: 1.2,
-      activeBuilders: 2,
+      medianTimeToCloseBugsHours: 1.2,
       projectsByProtocol: {
         spir: { count: 3, avgWallClockHours: 48.2 },
         bugfix: { count: 2, avgWallClockHours: 1.5 },
@@ -240,10 +239,9 @@ describe('AnalyticsView', () => {
     const stats = makeStats({
       activity: {
         prsMerged: 3,
-        avgTimeToMergeHours: null,
+        medianTimeToMergeHours: null,
         issuesClosed: 0,
-        avgTimeToCloseBugsHours: null,
-        activeBuilders: 0,
+        medianTimeToCloseBugsHours: null,
         projectsByProtocol: {},
       },
     });

diff --git a/packages/codev/dashboard/src/components/AnalyticsView.tsx b/packages/codev/dashboard/src/components/AnalyticsView.tsx
@@ -129,9 +129,8 @@ function ActivitySection({ activity, errors }: { activity: AnalyticsResponse['ac
       <MetricGrid>
         <Metric label="PRs Merged" value={String(activity.prsMerged)} />
         <Metric label="Issues Closed" value={String(activity.issuesClosed)} />
-        <Metric label="Avg Time to Merge" value={fmt(activity.avgTimeToMergeHours, 1, 'h')} />
-        <Metric label="Avg Time to Close Bugs" value={fmt(activity.avgTimeToCloseBugsHours, 1, 'h')} />
-        <Metric label="Active Builders" value={String(activity.activeBuilders)} />
+        <Metric label="Median Time to Merge" value={fmt(activity.medianTimeToMergeHours, 1, 'h')} />
+        <Metric label="Median Time to Close Bugs" value={fmt(activity.medianTimeToCloseBugsHours, 1, 'h')} />
       </MetricGrid>
     </Section>
   );

diff --git a/packages/codev/dashboard/src/lib/api.ts b/packages/codev/dashboard/src/lib/api.ts
@@ -152,10 +152,9 @@ export interface AnalyticsResponse {
   timeRange: '24h' | '7d' | '30d' | 'all';
   activity: {
     prsMerged: number;
-    avgTimeToMergeHours: number | null;
+    medianTimeToMergeHours: number | null;
     issuesClosed: number;
-    avgTimeToCloseBugsHours: number | null;
-    activeBuilders: number;
+    medianTimeToCloseBugsHours: number | null;
     projectsByProtocol: Record<string, ProtocolStats>;
   };
   consultation: {