DE{CODE} 2026 Developer Track

Build for the Web That Humans Browse and Machines Navigate

The web is running two systems at once. There’s the web your users browse, where design, performance, and experience determine whether they stay or leave. And then there’s the web AI agents crawl, interpret, and cite, where structured data, MCP compatibility, and semantic architecture determine whether your content even exists in AI-generated answers.

These four sessions from DE{CODE} 2026 are for the engineers and builders who have to master both. You’ll see live code, real demos, and working implementations. Learn about ACF’s MCP integration, a full headless AI tech stack built file by file, the conversion architecture for AI-influenced traffic, and the Abilities API, Connectors API, and PHP AI Client that enhance AI capabilities in WordPress® 7.0.¹

Pick the session most relevant to you or watch all four.

The ACF Update That Makes Your Content Machine-Readable

Runtime: 30:09

Your content is great, but AI can’t find it.

Rob Stinson and Iain Poulson go under the hood of ACF 6.8 and show how the plugin has evolved from a content modeling tool into essential infrastructure for the Intelligent Web. The decisions you make about structured data today will determine whether your content gets cited, or gets ignored, by AI-driven search.

Who This Is For: 

• Teams running ACF on any of the 5 million sites where it’s active
• WordPress developers managing custom fields and content architecture 
• Engineers preparing sites for AI crawlers and LLM citations 

What You’ll See and Build in this Session:

• How ACF 6.8 serves both human visitors and AI agents simultaneously without forcing you to choose between UX and machine readability.
• A live demo: Claude Code builds an entire custom post type, taxonomy structure, and field group from a CSV file and a plain-language prompt—zero manual UI clicks.
• Native, zero-code Schema.org mapping to 1,500+ properties directly in the ACF UI, with intelligent type detection for complex data (like prepTime for recipes) that generates valid JSON-LD automatically.
• What’s coming next: Smart Field Suggestions and WP-CLI commands for managing ACF JSON in CI/CD pipelines.

Fewer Clicks, Higher Stakes: How to Turn AI-Driven Traffic into Real Conversions

Runtime: 31:07

Organic human traffic is down. AI summaries are answering questions before users ever reach your site. But the visitors who do arrive are more informed and have higher intent than ever, and that changes everything about how your conversion layer needs to work.

Luke Patterson frames the architecture problem. Shane Daly demos the solution—Smart Search AI, AI-powered recommendations, and MCP integration from the WP Engine AI Toolkit—live, with working code.

Who This Is For: 

• Developers building or optimizing conversion architecture
• Engineers implementing search, recommendations, or chatbot features
• Teams who want to add MCP-grounded AI features without building custom middleware

What You’ll See and Build in this Session:

• Why 70% AI-influenced traffic changes your conversion architecture: Visitors arriving from LLMs aren’t browsing, they’re deciding. Your site needs to meet them there.
• Smart Search AI in action: Semantic search returns relevant results for concept-based queries (“financial misdeeds,” “justice”), even without exact-text matches, powered by fuzzy matching and vectorized content fields.
• The AI-powered recommendations block: The block drops into any Gutenberg theme in two clicks and surfaces related content by semantic similarity, not tags or categories, with no custom code required.
• Two ways to build an MCP-grounded chatbot: A developer-built RAG architecture using the AI SDK or a no-code setup using OpenAI’s agent builder—both are grounded in your content, both work seamlessly.

How Headless Gives You an AI Head Start

Runtime: 21:40

Who This Is For: 

• Developers evaluating or already building headless architectures 
• Engineers integrating AI agents, LLMs, or MCP servers 
• Teams building multichannel or high-scale AI applications

Headless WordPress isn’t just an architectural preference; it’s the prerequisite for modern AI integration.

Fran Agulto and Theresa Gobble demonstrate a full headless AI tech stack live: a WordPress Admin Copilot powered by Smart Search AI’s MCP server, a Next.js orchestration layer, and OpenAI, with every file open for inspection. If you want to understand how MCP, headless, and WordPress fit together in a real production architecture, this is the session to watch.

What You’ll See and Build in this Session:

• Why monolithic WordPress makes AI integration hard and exactly what headless unlocks: Clean REST and GraphQL APIs feed directly into LLMs, vector search, recommendation engines, and multichannel apps.
• The 2026 AI tooling landscape: Map the agentic workflows, context engineering, AI gateways, and sovereign small language models for regulated industries that are changing how sites operate.
• A file-by-file walkthrough of a real headless AI architecture: Learn about mcp.ts (MCP client integration), tools.ts (OpenAI tool-calling bridge), and route.ts (streaming API via server-sent events, with session recovery and iterative tool calling).
• Why headless handles AI-scale workloads better: Handle thousands of agent-generated landing pages per hour, hide database information behind API gateway and CDN, and dramatically reduce the DDoS and injection attack surface.

Featured Speakers:

• Francis Agulto: Developer Advocate, WP Engine
• Teresa Gobble: Software Engineer, WP Engine

WordPress 7.0 Is Here: What the Abilities API Actually Unlocks

Runtime: 28:49

7.0 didn’t arrive in isolation. Weston Ruter and John Parris, both principal software engineers at WP Engine and active WordPress core contributors, trace the arc from 6.9 to 7.0 and unpack the two most foundational shifts in the platform’s history: real-time collaborative editing and the new AI infrastructure stack.

If WordPress is becoming an AI context layer, this session shows you exactly how to build for it, including a live demo of a plugin being generated, installed, and activated from a plain-language prompt, without a zip file or FTP.

Who This Is For: 

• WordPress core developers and plugin authors 
• Engineers building AI-powered features inside the WP-admin 
• Anyone who wants to understand what the Abilities API, Connectors API, and PHP AI Client mean for how plugins get built

What You’ll See and Build in this Session:

• Real-time collaboration in WordPress 7.0: True co-editing is here, powered by Yjs CRDT, with WP Engine shipping WebSocket transport support at launch for high-traffic collaborative editing at scale.
• Why the Abilities API matters more than the REST API for AI: Use named, discoverable actions (“register a custom post type,” “publish a post”) rather than CRUD operations on objects—the true successor to Admin Ajax.
• The Connectors API in practice: One screen, one key, shared across all plugins, with built-in support for Anthropic (Claude), OpenAI, and Google Gemini, and a PHP AI Client that abstracts provider differences, so your plugins call one function regardless of the LLM underneath.
• Live demo: The AI Plugin Builder takes a plain-language prompt, generates a working plugin, and installs and activates it directly from the WP-admin—no zip file, no FTP, built by the WP Engine team at CloudFest Hackathon.

AI Infrastructure Is Here. What Are You Building with It?

These sessions are the technical foundation. WP Engine gives you the platform to build on it, from ACF’s structured data layer and the AI Toolkit to headless architecture and adaptations to harness the full functionality of WordPress 7.0’s AI infrastructure stack. Whether you’re integrating MCP, building for agentic traffic, or shipping the next generation of plugins, we’re built for what you’re building next.