New strategy post

content/news/2026-06-06-the-ai-development-boom-is-great.md

@@ -0,0 +1,62 @@
+---
+date: 2026-06-06
+author: Aaron Hans
+title: "The AI development boom is great. Automated quality checks make it sustainable."
+description: "AI-generated code has consistent quality gaps across accessibility, security, and bot-friendliness. Automated guardrails can catch them before they ship."
+eleventyExcludeFromCollections: true
+eleventyComputed:
+  permalink: /news/ai-development-boom-automated-quality-checks/
+keywords:
+ - AI code quality
+ - web standards
+ - accessibility
+ - automated testing
+---
+
+More people can ship software now than ever before. AI development tools let content creators, program managers, and domain experts build pages and web applications directly. That's a good thing. The people who understand the problems best can now build the solutions.
+
+But AI-generated code has consistent quality gaps.
+
+## What we're seeing
+
+ScanGov audits thousands of websites for bot-friendliness, accessibility, usability, and security. The same failures show up in AI-generated code:
+
+- **Accessibility gaps are the default output.** AI-generated HTML misses semantic elements, breaks heading order, and skips ARIA attributes. These aren't edge cases. In our audits, broken headings and missing link names are among the most common failures — and AI tools produce both reliably.
+
+- **Client-side rendering hides content.** AI tools often generate JavaScript-heavy solutions without flagging the tradeoffs. The page works in a browser. But any bot (like AI crawlers) that doesn't run JavaScript sees an empty shell. We measure this by comparing what the server sends against what appears after JavaScript runs. The gap is often large.
+
+- **Performance degrades.** Extra dependencies and unnecessary JavaScript add up. [90% of government websites already fail performance thresholds](https://www.linkedin.com/pulse/90-government-websites-have-performance-issues-aaron-hans-m8zpc/). The main cause is too many client-side resources, not slow servers. AI-generated code makes this worse.
+
+- **Security headers get missed.** CSP policies, HSTS, HTTPS — these are invisible to users but critical for safety. Non-specialists don't think to check them. AI tools don't reliably add them.
+
+These problems aren't unique to one AI tool. The models produce code that looks right and works in a browser but fails standards checks.
+
+## Scanning regularly isn't enough
+
+Traditional monitoring assumes a small team ships changes. Someone runs a scan. Someone reviews the report. Fixes get scheduled for the next sprint.
+
+That made sense when a few developers controlled what went live. It doesn't work when twenty people across an organization can ship changes in an afternoon. By the time a quarterly scan catches a problem, the page has been live for weeks.
+
+## Guardrails in the development workflow
+
+ScanGov can check code against web standards before it ships as well as monitor your production environment.
+
+- **CI pipeline.** Every pull request gets checked automatically. Accessibility, rendering, performance, and security issues are flagged before code reaches production. We render pages in a real browser, so we catch problems that static HTML scanners miss.
+
+- **MCP server for AI tools.** Connect ScanGov to Claude or other AI assistants. When someone builds a page with AI, ScanGov checks their work in the same tool. The AI gets feedback on what to fix before a commit happens.
+
+- **Editor integration.** Run checks locally as you build. Get results without waiting for CI.
+
+- **API.** Add standards checking to any workflow — CMS, design tools, custom builds.
+
+- **Live site monitoring.** Continuous scanning catches what slips through and tracks trends over time across all your sites.
+
+## What we check that others don't
+
+Popular audit tools focus on performance and accessibility. That's important, but it's not the full picture.
+
+ScanGov also checks whether bots can read your content, whether your writing meets plain language guidelines, and whether your security headers follow federal standards. We started from government web requirements and expanded into usability and bot-friendliness checks that general-purpose tools don't cover.
+
+## See where your site stands
+
+Scan any site for free at [scangov.com](https://scangov.com). You'll get results across bot-friendliness, accessibility, usability, and security in seconds.

public/css/scangov.css

@@ -105,13 +105,16 @@
     --bs-form-control-focus-color: var(--bs-body-color);
     --bs-card-focus-box-shadow: 0 0 0 0.25rem rgba(112, 225, 123, 0.25);
     /* Status colors */
+    --bs-success: #70e17b;
     --bs-bg-success: #70e17b;
     --bs-bg-success-rgb: 112, 225, 123;
     --bs-border-success: #70e17b;
+    --bs-warning: #face00;
     --bs-text-warning: #000000;
     --bs-bg-warning: #face00;
     --bs-bg-warning-rgb: 250, 206, 0;
     --bs-border-warning: #face00;
+    --bs-danger: #e41d3d;
     --bs-text-danger: #ffffff;
     --bs-bg-danger: #e41d3d;
     --bs-bg-danger-rgb: 228, 29, 61;
@@ -1837,4 +1840,12 @@ th a:hover i, th a:hover svg,
 th a:focus i, th a:focus svg {
     color: var(--bs-body-color) !important;
     fill: var(--bs-body-color) !important;
-}
+}
+/* Docs */
+#post h3 {
+    margin-top: 2rem;
+}
+
+.js-sidenav a:hover {
+    color: var(--bs-link-color) !important;
+}