OpenStation vs Smart Local AI – Browser-Based Private AI Tools

wp-admin forgets. OpenStation doesn’t. You know the moment. You’re deep in a post and you need one image from the Media Library. You click over, find it, click back, and everything you had going is gone. Your scroll position, your half-typed thought, the three other things you had queued up in your head. That’s not a bug you hit. That’s the design. wp-admin shows you one screen at a time, and every navigation throws the last one away. It has worked this way since 2003, like a fridge that empties itself every time you close the door. WordPress has rebuilt almost everything else. The editor, the site editor, the patterns, the styles. The admin’s one-screen-at-a-time model is the layer nobody touched. OpenStation touches it. Install it and admin screens become draggable, resizable windows. Edit a post while the Media Library sits open next to it. Drag an image from one window straight into the other. Keep comments open in the corner. Reload the browser and your window layout — open pages, positions, sizes, and window states — comes back where you left it. OpenStation itself is free, GPL, and opt-in per user. One click in the admin bar turns it on for you. One click turns it off. Nobody else on your site sees anything change, and deactivating the plugin restores the classic admin exactly. No Core patches and no lock-in. Optional AI providers may charge for their service. Built and maintained by Automattic, the company behind WordPress.com, Jetpack, WooCommerce, and Tumblr. Work on more than one thing at once Admin pages open as windows, including pages added by most other plugins. Drag them, resize them, snap them, tile them, minimize them to a taskbar. The admin menu becomes a dock of icons. Two editors open at once is one click away. Never lose your place Moving between open windows leaves each page in place. Across reloads, OpenStation remembers your open pages, window positions, sizes and states, virtual desktops, and focus. Closing a window with unsaved changes warns you first. A desk you can actually put things on Drop posts, media, and links right onto the wallpaper. Sort them into folders. Trash them to a Trash that actually holds things. The stuff you’re working on this week can just sit out where you can see it. One desktop per job Virtual desktops, the way your operating system does them. One for writing. One for the store. One for comment triage. One for that redesign you keep poking at. Switch between them without closing anything. Find anything with Cmd+K The full WordPress command palette, plus slash commands from your plugins, all under one keystroke. Turn on the optional AI assistant and you can ask it things like “which post had the comment asking for the recipe?” and it searches your own content to answer. (Requires an AI provider configured in Settings → Connectors on WordPress 7.0+. Details under “External services” below.) See how your site fits together The Corkboard is an interactive, zoomable map of how your posts, pages, and products link to each other. Pan around it. Find the orphans. Find the hubs. Running a store? Orders, Products, Coupons, and Customers become browsable folders with stock and sale ribbons. Open a customer to see their lifetime spend and what they buy most. Follow a product to its buyers, or a coupon to who redeemed it. Make it yours Wallpapers, from color presets to animated scenes to your own photo. Live widgets pinned to the desktop: drafts, stats, recent comments, a focus timer. Uploadable desktop themes, including a built-in Legacy theme that brings the old WordPress blue back in one click. And Mio, a soft-body desk companion that drifts across your wallpaper and gets bumped around by your windows. Off by default. You’ll turn it on. For developers Extend OpenStation through documented PHP and JavaScript APIs. Register windows, icons, wallpapers, widgets, commands, and settings tabs through stable openstation_register_* PHP APIs and a typed JavaScript API; customize dock items through documented hooks; add AI tools with the WordPress Abilities API. Copy-paste examples live in the developer docs on GitHub. External services No external service is required for OpenStation’s desktop interface. The optional AI Assistant and two user-initiated enrichment features make the external requests described below. AI Assistant The optional AI Assistant sends data to the AI provider you configure in WordPress’s Settings → Connectors (for example OpenAI, Anthropic, or Google). Generation is routed through WordPress 7.0’s built-in AI Client, which supplies the credentials stored in Connectors. The plugin never handles an API key itself. With no provider configured in Connectors, no external AI requests are made. When the AI Assistant is enabled and a user invokes it (via Cmd+K or the slash-command palette): What is sent: the user’s prompt, the conversation history for the active session, and tool-call metadata. The plugin’s built-in tools (search_posts, search_pages, search_comments) run WordPress’s native keyword search and may include excerpts of the matching posts/pages/comments in tool results, which are then sent back to the provider as part of the agentic loop. When it is sent: on user-initiated AI requests, and (if an administrator enables “Score new comments with AI”) on comment-save hooks for spam analysis. Posts, pages, and taxonomy terms are not sent automatically. Why it is sent: to obtain model completions and tool-call decisions that drive the AI Assistant. Who provides the service: whichever provider you configured in Settings → Connectors. Which provider (and endpoint) receives the data depends entirely on that configuration. Review the chosen provider’s own terms and privacy policy (e.g. OpenAI, Anthropic, or Google). URL shortcut favicons When an authorized user creates a desktop shortcut to an external URL, OpenStation asks that URL for its page HTML and favicon so the shortcut can display the site’s icon. The request is made from your WordPress server and sends the requested URL, the server’s IP address, an OpenStation user-agent string, and normal HTTP request metadata to the operator of that site. This happens only when a user creates the shortcut. The destination site’s terms and privacy policy apply. WordPress.org plugin information When an authorized user opens or refreshes parts of OpenStation’s Plugins window, OpenStation may request public plugin details, update metadata, and review excerpts from WordPress.org. These requests send plugin slugs, the site’s locale, and normal HTTP request metadata to WordPress.org. This information is used only to display and manage plugins. Review the WordPress.org Privacy Policy. OpenStation’s OAuth relay does not contact a service by itself. Third-party plugins that configure the relay may contact their own providers and are responsible for disclosing those services. Credits OpenStation is brought to you by Automattic. The plugin is open source under GPLv2-or-later; contributions are welcome on GitHub. Third-party libraries The plugin bundles the following third-party JavaScript library, loaded on demand only when a feature that needs it is in use: PixiJS (MIT License) — used by the Corkboard window, built-in canvas wallpapers (e.g. the animated WordPress logo), the Mio desk companion, and the Inkfall typing game. PixiJS is loaded from the plugin’s own assets/vendor/ directory; no CDN requests are made. Data files The Inkfall game’s word list (assets/games/inkfall/words.txt) is generated from the following sources (attribution also ships in the file’s header): FrequencyWords by Hermit Dave (CC-BY-SA 4.0) — English word-frequency ranking derived from the OpenSubtitles corpus. english-words by dwyl (Unlicense) — used as a validity filter. LDNOOBW English list (CC-BY 4.0) — used as an exclusion filter.

Smart Local AI brings machine learning directly to the browser. Every ML model runs on-device in the visitor’s or editor’s browser using Transformers.js and WebGPU/WASM. No data is ever sent to external servers, no cloud API keys are needed, and there are zero recurring costs. Modules RelevantFlow — Semantic Content Recommendations Traditional related-posts plugins match by shared tags or categories. RelevantFlow uses a sentence-transformer model (all-MiniLM-L6-v2) to understand the meaning of your content and recommend posts that are genuinely related — even when they share no tags at all. Understands synonyms and concepts (e.g. “Salah” and “Islamic Prayer” match) Works with any content type — Posts, Pages, WooCommerce Products, or custom post types Displays a responsive card widget below your content Three layout templates: cards, list, and grid Custom placement via CSS selector or shortcode [atlas_ai_relevantflow] Theme customization with presets (Default, Light, Dark, Minimal) and custom colors Embedding index built in-browser by editors — zero server load Nightly cron flags new/updated content for re-indexing PersonaFlow — Personalized Content Recommendations PersonaFlow delivers personalized content recommendations based on visitor behavior — like Netflix for your WordPress site. It tracks 30 behavioral signals (scroll depth, reading time, click patterns, etc.) and computes a per-visitor taste vector to surface the most relevant content for each individual visitor. Personalized recommendations powered by real-time behavioral tracking 30 behavioral signals including scroll depth, reading time, and engagement patterns Per-visitor taste profile computed entirely in the browser — no server-side tracking WooCommerce integration for product recommendations Three layout templates: cards, list, and grid Custom placement via CSS selector or shortcode [atlas_ai_personaflow] Theme customization with presets and custom colors All tracking data stays on the visitor’s device — fully GDPR compliant Automatic taste decay ensures recommendations stay fresh and relevant AltGenius — Automatic Alt Text Generation Missing alt text hurts SEO rankings and violates ADA/WCAG accessibility guidelines. AltGenius generates descriptive alt text for every image in your media library — entirely in the browser. Two AI models: ViT-GPT2 (fast) and Florence-2 (detailed) Bulk Scan to audit and fix your entire media library at once Auto-generate on media upload Editor sidebar panel for per-image control with prefix/suffix No per-image fees, no cloud processing Key Features Zero Cost — No API keys, no subscriptions, no per-request fees Privacy by Architecture — All ML inference runs in the browser; data never leaves the device GDPR Compliant — No external data transmission means no DPA required Shared Hosting Compatible — No server-side ML dependencies; works anywhere WordPress runs WebGPU + WASM — Hardware-accelerated inference with automatic fallback Modular — Enable only the modules you need; each can work standalone Cached Models — ML models are cached in IndexedDB after first download for instant reuse Themeable — Built-in theme presets and custom color options to match any WordPress theme Upgrade to Pro Get more powerful AI modules and premium features with Smart Local AI Pro. Have questions or need support? Contact us. How It Works All AI processing runs in your browser. No server-side ML is required. On first use, models are downloaded from Hugging Face and cached in the browser’s IndexedDB — this one-time download may take 1–2 minutes depending on your connection. After that, models load instantly. RelevantFlow (Related Posts) Go to Smart Local AI > Dashboard and enable the RelevantFlow module (or follow the setup wizard after activation) Select which content type to index from the Content Type selector on the Dashboard (defaults to Posts) Go to the RelevantFlow tab and click Build Index — this downloads the MiniLM model (~23 MB) and generates an embedding for each published item of the selected content type You need at least 2 published items of the selected content type for recommendations to appear Visit any item on the frontend — a “You Might Also Enjoy” section will appear below the content showing semantically related items PersonaFlow (Personalized Recommendations) Go to Smart Local AI > Dashboard and enable the PersonaFlow module Go to the PersonaFlow tab to configure display settings (number of recommendations, layout template, appearance) Visit posts on the frontend — PersonaFlow will track reading behavior and build a taste profile After a few page views, personalized recommendations will appear below the content based on the visitor’s interests Use the shortcode [atlas_ai_personaflow] to place recommendations anywhere in your theme AltGenius (Image Alt Text) Go to Smart Local AI > Dashboard and enable the AltGenius module Upload an image via Media > Add New Media File — the model downloads on first use (~300 MB for ViT-GPT2) and alt text is generated automatically Or go to Smart Local AI > Bulk Alt Text and click Start Bulk Scan to generate alt text for all existing images missing it Or edit any post with images in the Block Editor — the AltGenius sidebar panel shows each image with a Generate Alt Text button External Services This plugin connects to the following external services. No user data or site content is transmitted to these services — only the ML model files are downloaded. Hugging Face Model Hub ML models are downloaded from the Hugging Face Model Hub (https://huggingface.co/) on first use and cached in the browser’s IndexedDB. The following models are used: RelevantFlow: Xenova/all-MiniLM-L6-v2 (~23 MB) — Sentence transformer for content embeddings AltGenius (ViT-GPT2): Xenova/vit-gpt2-image-captioning (~300 MB) — Image captioning model AltGenius (Florence-2): onnx-community/Florence-2-base-ft (~460 MB) — Advanced image captioning model These models are open-source and licensed under Apache 2.0 or MIT. No user content, images, or site data is uploaded to Hugging Face — the connection is download-only. Hugging Face Terms of Service Hugging Face Privacy Policy