WordPress 7: What’s next for the leading CMS?
WordPress 7 marks one of the most exciting developments in recent years. In particular, the integration of AI features is taking center stage and opening up new possibilities for everyone working with WordPress.
While many users assess the new features primarily in terms of usability, developers often take a more nuanced view of the changes in terms of architecture, extensibility, and long-term viability.
Businesses and website owners are also asking whether WordPress will remain a reliable, future-proof platform for their digital presence.
I’ll take a look at the most important new features and assess the opportunities and challenges that the new WordPress brings.
In any case, these developments offer hope that WordPress will remain agile enough to keep pace with the rapid evolution of the internet.
Why use WordPress at all?
It’s a perfectly valid question. There are important reasons why it still makes sense to use WordPress as a foundation.
WordPress is the right choice for
a sustainable, maintainable solution that can also be managed by non-developers.
Vibe coding / custom development is ideal for
single landing pages, small tools (e.g. calculators), or dashboards that do not require access management (user accounts). Automatically generating a single web page from a PDF or video is also an exceptionally cost-effective solution.
Cost structure:
AI cloud services charge based on usage, while WordPress offers a much more predictable and cost-effective foundation. Hosting under your own domain incurs costs in either case.
Examples:
Return calculator for the events industry
StageAID return calculator
| Criterion | WordPress (CMS ecosystem) |
Vibe-coded / custom-built (HTML/CSS/JS/PHP) |
|---|---|---|
| Content management system (backend) | Built-in editor Ready-to-use admin dashboard, including an editor for pages and a media library. |
No backend Content must be edited directly in the code, possibly by entering a new prompt. |
| Extensibility & plugins | One click away More than 60,000 ready-made plugins (shops, forms, SEO, bookings) can be installed immediately. |
Manual work Every feature and API integration must be individually developed and maintained in code. |
| Maintenance & security | Established standard Security updates for core, themes, and plugins can be installed automatically. |
Your responsibility No automatic patching or built-in GDPR compliance. AI-generated code may contain vulnerabilities that you must secure yourself. |
| SEO & marketing | Guided visually Plugins (Rank Math, Yoast) provide structured data, sitemaps, and content analysis without code. |
Manual HTML Meta tags, Schema.org, and Open Graph must be maintained precisely in the code. |
| Roles & permissions | Built in Roles such as administrator, editor, author, and customer come with a complete permission system. |
Complex Authentication, password resets, and permission checks must be implemented securely. |
| Future-proofing & handover | Industry standard Almost any developer or designer can take over and continue a WordPress project. |
Code vendor lock-in In some cases, projects cannot easily be moved to another provider. |
| Performance & resources | More overhead Requires a server with PHP/database capacity and often loads unused code through themes and plugins. |
Extremely fast Lean, pure code without database overhead delivers maximum speed and low server load. |
The current state of the art, briefly explained.
One important point first:
WordPress 7 does not include its own artificial intelligence.
Instead, the system provides a standardized AI interface that plugins and themes can use to communicate with different AI services.
This abstraction layer supports OpenAI (GPT), Anthropic (Claude), Google (Gemini), xAI (Grok), Mistral, and Perplexity, as well as local models such as Ollama through additional extensions.
Developers use a unified API in PHP or JavaScript, or access it through the WordPress REST API, regardless of which AI provider is actually being used.
Categories of AI features
| Category | Description | Typical use cases |
|---|---|---|
| Content & editing | AI supports the creation, editing, and optimization of content in day-to-day operations. | Writing, image generation, translations, SEO optimization, summaries, and more. |
| Development & maintenance | AI supports developers and administrators in creating, maintaining, and optimizing WordPress projects. | Code generation, debugging, refactoring, boilerplate code, documentation, database queries, test cases |
| AI applications & automation | AI becomes an integral part of the website or internal business processes and is made available to visitors or employees. | Chatbots, AI agents, knowledge bases (RAG), intelligent search, product recommendations, form assistants, workflow automation |
Setting up the interface with an API key takes just a minute, provided you already have an account with one of the platforms. However, this alone does not activate any new features in the WordPress backend. The AI interface only provides the technical foundation for plugins or custom extensions to use. Choosing the right plugin is therefore essential. Below, I focus on three well-executed solutions with AI integration.
WordPress plugins with interesting AI features
| Plugin | Use cases |
|---|---|
|
Novamira (Getting started with Novamira) Category: Development & maintenance, Content & editing |
Connects WordPress directly to AI agents such as Claude Code, Cursor, or GitHub Copilot via the Model Context Protocol (MCP). The AI can execute PHP code, edit files, run database queries, manage plugins, and handle complex development tasks directly in a development or staging environment. |
|
AI Engine by MeowApps
Category: AI applications & automation, development & maintenance |
A universal AI framework for WordPress. It creates text and images, and provides chatbots, forms, translations, and automation. It supports numerous AI providers such as OpenAI, Anthropic, and Gemini. Its own API and MCP support make it suitable both for content editors and for developing custom AI applications. |
| LiveCanvas AI (Website)
Category: Development & maintenance |
AI assistant for visual website building. It helps create HTML and optimize code, and offers a screenshot-to-code feature designed to automatically convert designs into Bootstrap 5-compatible HTML. The generated elements can then be edited directly in the LiveCanvas editor. |
Topic: Page builders and block editors
Elementor, Divi, Avada, WPBakery, Gutenberg...
They all now include an AI chat that is intended to let you create custom layouts and pages from prompts. As a result, they mostly belong in the "Content & editing" category. Personally, I find the implementation in Elementor, for example, rather clunky and imprecise. The PRO version also incurs additional costs.
A lightweight and agile alternative to all those bloated page builders is livecanvas
A particularly interesting example is the image-to-code feature.
A design or screenshot is uploaded as an image and then automatically converted into HTML and CSS. Ideally, the AI generates Bootstrap 5-compatible code that can be inserted directly into the editor.
With LiveCanvas, this means the generated elements are immediately available as containers, rows, columns, or Bootstrap components.
AI features can also benefit your website visitors.
AI applications can also be built with WordPress.
For example, an intelligent document analysis tool or an AI shopping assistant.
WordPress + AI plugins are ideal for
Businesses, editorial teams, and content teams that want to use ready-made AI tools (content generation, chatbots, SEO optimization) in their daily work without developer expertise.
Cloud-native AI is the right choice for
custom SaaS applications, high-performance AI agents, tailored RAG pipelines, and products that require full model control and maximum development freedom.
Cost structure:
Cloud-native services (Vercel/OpenAI) incur ongoing, usage-based API token fees, while WordPress offers a much more predictable and budget-friendly foundation.
| Criterion | WordPress + AI plugins | Cloud-native AI (Vercel / Google AI / OpenAI platform) |
|---|---|---|
| Architecture & setup |
Integrated with the CMS AI plugins use the WordPress database and integrate directly with the Gutenberg editor and backend. |
Custom code / API A custom frontend/backend app (e.g. Next.js on Vercel) connected directly to AI models via SDKs. |
| Target users & usability |
No-code / content editors Simple, ready-made interfaces for content managers, marketers, and administrators without programming skills. |
Developers & tech teams Requires expertise in JavaScript/TypeScript, Python, serverless endpoints, and API state management. |
| Use cases |
Content & business websites Automated blogging, SEO copywriting, image generation, product descriptions, and customer support chatbots. |
AI products & SaaS Standalone web apps, complex RAG systems, autonomous AI agents, voice AI, and real-time inference. |
| Scalability & speed |
Monolithic AI processes run in the PHP stack; parallel synchronous AI calls can affect server performance. |
Unlimited & edge Serverless architecture enables fast execution, asynchronous workflows, and unlimited concurrent users. |
| Costs & billing |
Predictable & manageable Fixed license fees for plugins and standard hosting (API token costs usually remain in the cents range). |
Usage-based (pay as you go) Billed per token/request (OpenAI/Google), plus edge computing costs (Vercel). |
WordPress with its own local AI platform (on-premises AI environment)
In this scenario, the AI infrastructure runs entirely on-premises.
In my case, the WordPress development environment runs on a workstation using WP Local (a local WordPress server). In addition, an Ubuntu VM runs in Parallels Desktop (16 GB RAM / 4 processor cores), hosting a dedicated AI platform with Ollama and local language models.
The Ubuntu VM provides AI services through an internal API. WordPress communicates with this AI interface over HTTP/REST and can therefore use local language models for a range of tasks.
WordPress + a local LLM is ideal for
privacy-sensitive businesses, intranets, law firms, medical institutions, or government agencies that need to operate in compliance with the GDPR without external cloud APIs.
Hardware & servers
Standard shared hosting is not sufficient for local models. You need a server with a dedicated GPU or a separate computer on the local network running Ollama or LM Studio.
One-time investment vs. API fees: Setup requires powerful hardware, but the reward is €0 in ongoing token costs, no matter how extensive the usage.
| Criterion | WordPress + local LLM (e.g. Ollama) | WordPress + cloud LLM (OpenAI / Claude) |
|---|---|---|
| Data privacy & GDPR |
100% on-premises No data leaves the system. Maximum protection for confidential content, intranets, and customer data. |
Cloud transfer Prompts are sent to US/EU servers. Data processing agreements and privacy disclosures are essential. |
| Ongoing AI costs |
€0 in token costs No limits or charges per generation or word. Ideal for high-volume content and extensive testing. |
Pay per token Usage-based billing per prompt, token, and vector embedding (costs can rise as you scale). |
| Hosting & hardware |
GPU required Requires a local GPU server (e.g. NVIDIA VRAM) or Ollama running on a Mac/PC on the same network. |
Minimal hosting Standard PHP web hosting is sufficient because computation takes place externally with the cloud provider. |
| Response speed |
Hardware-dependent Very fast on a powerful GPU; extremely slow on server CPUs alone or with insufficient VRAM. |
Consistent & fast Large data centers ensure fast response times (e.g. GPT-4o Mini / Gemini Flash). |
| Setup & integration |
Moderate effort Install Ollama/LM Studio and enter the API endpoint (e.g. http://127.0.0.1:11434) in the WordPress plugin. |
Plug & play Copy the provider’s API key, paste it into the WordPress plugin, and use all features right away. |
WordPress & a local LLM on your own hardware.
Current use case: Smartphone Attack Vector
The database has already collected more than 50 different attack vectors.
"Identifying and defending against attacks on smartphones over mobile networks".
The site analyzes ways to attack mobile devices. It examines the technical vulnerabilities, attack paths, and security risks found in smartphones, networks, operating systems, apps, and communication interfaces.
The site is currently being updated with information about 4G and 5G mobile networks.
The local AI environment is intended not only to create new content, but also to serve as a continuous quality assurance and update system for the database.
Hardware & software: An Ubuntu VM with 16 GB RAM, the Ollama LLM runtime, Qwen, and DeepSeek.
Deployment: Once a new post is complete, the final result is deployed to the live server.
The live site runs a standard WordPress installation and does not require any AI interfaces.
I hope this brief overview has shed some light on the subject.
Do you have any questions, suggestions, or feedback? I’d be happy to hear from you!
Just send me an email at:
info@dm-development.de
- This text was translated by AI -