This was the year AI became part of the everyday internet. It helped teams design faster, code smarter, launch tools with less friction, and personalize digital experiences in ways that once seemed out of reach. At the same time, it raised important questions about trust, structure, ethics, and where people still fit in the loop.
Across our coverage, a few themes emerged. From the evolution of WordPress as an AI-native platform to the infrastructure powering AI at scale, these were the ideas and implementations that shaped 2025. Below, we’ve grouped the most discussed and most referenced posts by focus area, each offering a snapshot of how teams are approaching AI with strategy and purpose.
🧱 WordPress as an AI-Ready Platform
How AI Is Reshaping WordPress Block Development
AI is changing how developers prototype, scaffold, and deploy WordPress features. This post explored tools like Telex, WP-AutoPlugin, WPCode’s AI Snippet Generator, and QuickDeployWP. Whether generating blocks from a prompt or building entire plugin folders from scratch, these tools are speeding up development and freeing up teams to focus on high-value tasks. We also explored how model-agnostic tooling gives teams more flexibility and how real-world developers are pairing AI with their existing workflows.
AI Building Blocks Have Arrived to Transform What’s Possible with WordPress
Beyond standalone tools, WordPress core is laying the groundwork for AI-native development. This post covered the three key initiatives launched by the Make WordPress AI team: the PHP AI SDK for model interoperability, the Abilities API to expose WordPress capabilities to AI agents, and the MCP Adapter to align with emerging AI protocols. Together, these efforts are setting a foundation for long-term integrations that can support personalization, automation, and conversational interfaces inside WordPress environments.
Making AI Work in WordPress: Lessons from DE{CODE} 2025
This post recaps our full DE{CODE} 2025 session with WP Engine’s Luke Patterson and Elastic’s Justin Castilla, where we moved past buzzwords and showed what AI looks like when it is grounded in real data, real architecture, and real editorial needs. The post walks through Elastic’s breakdown of vector stores, hybrid search, and RAG pipelines, then shifts into Trew Knowledge’s showcase of Edutopia’s AI-powered ecosystem, led by Jeffrey Zalischi and Anthony Moore. Readers get an inside look at how we started with a lightweight generative tool, then evolved it into a WordPress‑integrated editorial assistant, a custom Gutenberg block, a vectorized content knowledge base, and a personalization framework powered by Neo4j. The session demonstrated how AI can support editors, strengthen content discovery, and create meaningful user experiences when the foundation is right.
🔍 Discovery and Personalization, Reimagined
Revolutionizing Content Discovery Through AI
Built by Trew Knowledge for Edutopia, this assistant introduced a new kind of search experience. Instead of keyword matches, it uses OpenAI models and a private vector database to interpret educator questions and return context-aware answers based on verified content. The post walked through how the system understands nuances like “disruptive behaviour in math class” and links it to relevant topics like inquiry-based learning or classroom routines. This was a real-world example of what happens when AI is paired with curated content and clear boundaries.
Personalization in Action: The Architecture Behind Edutopia Homeroom
Homeroom took personalization further by creating a dynamic content feed that evolves with each user’s activity. Powered by vector embeddings and a graph database, the system maps relationships between content, users, and topics. The post explained how these connections form organically, enabling recommendations based on concept proximity rather than metadata tags. We also detailed how Neo4j handles the interaction graph, and how this model supports user engagement, content discovery, and professional learning at scale.
🎨 AI in Design, Content, and Workflows
AI in Creative Workflows: Our Hands-On Tests with Nano Banana
We tested Google’s Gemini-powered image model, Nano Banana, across a range of creative use cases: moodboards, product shots, headshot editing, brand consistency, and 3D renders. This post shared the results from real experiments, highlighting what worked well and where human oversight was still critical. The model’s ability to maintain logo placement, iterate on prompts, and style-match across compositions impressed our team. But it also raised important questions about creative sameness, visual realism, and the role of curation in fast-moving campaigns.
🛠️ Enterprise AI Infrastructure and Scale
OpenAI DevDay 2025: The Year Chat Became a Platform
This post unpacks how OpenAI shifted from an API provider to a full ecosystem platform. Highlights included the ChatGPT Apps SDK, AgentKit for building AI agents with drag-and-drop logic, and new models like GPT-5 Pro and Sora 2 for voice, code, and video generation. We also explored the rise of connectors, guardrails, and monetization options inside ChatGPT, showing how developers are turning conversational interfaces into end-to-end user experiences. The framing centred around flexibility, control, and safety at scale.
AWS re:Invent 2025: When Enterprise AI Grew Up
While OpenAI focused on interaction, AWS focused on infrastructure. This post covered the Nova 2 model family, including real-time voice, multimodal reasoning, and fine-tuning via Nova Forge. We looked at AgentCore for building autonomous agents, vector embedding support in S3, and new GPU instances built for massive training workloads. Durable Functions in Lambda signaled a shift toward long-running AI workflows, while the concept of “Novellas” let teams extend base models without starting from scratch. The post outlined what it means for large organizations to move from pilot projects to production-ready AI systems.
🌐 Global Trends and Human-Centred AI
Canada’s Rise as a Global AI Hub
This post tracked how cities like Toronto, Waterloo, and Vancouver emerged as top-tier tech markets, thanks to strong university pipelines, global talent access, and early investment in AI research. We looked at how government policy, academic partnerships, and immigration strategy combined to create a stable, high-growth ecosystem. The story connected macroeconomic shifts to platform decisions and explained why more enterprises are choosing to expand or relocate their AI operations to Canada.
The Human Factor in an AI-Driven World
This reflection piece tackled the most important question of the year: what’s left for people? From design judgment and brand identity to oversight, originality, and cultural awareness, we explored the irreplaceable value of human input. The post also examined regulation trends, like the EU AI Act and North American guidance, and broke down practical considerations for content authorship, data privacy, and ethical review. AI can generate, but people still shape the story. That’s what keeps it useful, respectful, and real.
What 2025 Showed Us About AI
If there’s one thing these posts made clear, it’s that AI isn’t a separate track anymore. It’s embedded in the tools we use, the workflows we trust, and the decisions we make, whether we’re building for education, publishing at scale, or running enterprise platforms.
But it’s not just about what AI can do. It’s about how we structure content, how we set guardrails, and how we decide what gets automated, and what doesn’t. The teams making real progress are the ones pairing innovation with intent, weaving AI into places where it adds clarity, speed, or personalization, and keeping humans in control of the outcomes.
These conversations will keep evolving. And so will the tools, models, and questions that come with them.
To keep up with it all, we publish AI This Week, a series that tracks the launches, breakthroughs, and practical insights shaping the AI landscape. If you’ve been following along, we appreciate the time and curiosity. If you’re just tuning in, now’s a great moment to explore what you’ve missed.
We’ll be back when the next wave of updates hits — and it won’t take long.
