Smart Local AI – Browser-Based Private AI Tools

Description

Smart Local AI brings machine learning directly to the browser. Every ML model runs on-device in your (the editor’s) browser using Transformers.js and WebGPU/WASM. Your posts and images are never uploaded to an AI service, no cloud API keys are needed, and there are zero recurring costs. The only AI-related outside request is the one-time model download from Hugging Face, which you start yourself (see “External Services”).

A setup wizard opens on your first visit to Smart Local AI. It checks what your browser can run, shows the size and source of each model before you download it, and lets you try related posts and alt text on your own content before anything is saved.

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 on your own server from the recorded signals
• 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
• Tracking data is stored in your own WordPress database — never sent to us or to a third party (mention it in your privacy policy)
• 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; your content and images are never uploaded to an AI service
• No third-party processing — Recommendations and personalization data stay in your own WordPress database
• 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 kept in the browser’s Cache Storage after the first download for instant reuse (browsers only keep them on HTTPS sites)
• 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. Models are downloaded from Hugging Face only when you start the download (in the setup wizard, or when you first use a module) and are then kept in the browser’s Cache Storage — this one-time download may take 1–2 minutes depending on your connection. After that, models load instantly.

RelevantFlow (Related Posts)

1. Go to Smart Local AI > Dashboard and enable the RelevantFlow module (or follow the setup wizard after activation)
2. Select which content type to index from the Content Type selector on the Dashboard (defaults to Posts)
3. 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
4. You need at least 2 published items of the selected content type for recommendations to appear
5. Visit any item on the frontend — a “You Might Also Enjoy” section will appear below the content showing semantically related items

PersonaFlow (Personalized Recommendations)

1. Go to Smart Local AI > Dashboard and enable the PersonaFlow module
2. Go to the PersonaFlow tab to configure display settings (number of recommendations, layout template, appearance)
3. Visit posts on the frontend — PersonaFlow will track reading behavior and build a taste profile
4. After a few page views, personalized recommendations will appear below the content based on the visitor’s interests
5. Use the shortcode [atlas_ai_personaflow] to place recommendations anywhere in your theme

AltGenius (Image Alt Text)

1. Go to Smart Local AI > Dashboard and enable the AltGenius module
2. 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
3. Or go to Smart Local AI > Bulk Alt Text and click Start Bulk Scan to generate alt text for all existing images missing it
4. 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. Your posts, images and visitor data are never sent to any of them.

Hugging Face Model Hub

ML models are downloaded from the Hugging Face Model Hub (https://huggingface.co/ and its file download servers) by your browser, only when you start a download (in the setup wizard, the module settings, or when you first use a module), and are then kept in the browser’s Cache Storage. Like any download, Hugging Face sees the downloading browser’s IP address. If you enter a Hugging Face access token for Florence-2, it is sent to Hugging Face with the model download. The ONNX Runtime WebAssembly engine that runs the models ships with the plugin and is not downloaded from a CDN. 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 (~250 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

AtlasAiDev plugin list and WordPress.org plugin information

When an administrator opens Smart Local AI > Plugins, your server downloads the list of AtlasAiDev plugins from GitHub (https://raw.githubusercontent.com/atlasaidev/plugins/main/plugins.json) and their ratings and install counts from the WordPress.org Plugins API (https://api.wordpress.org/plugins/info/1.2/). Both are cached for 24 hours. No site data is sent.

• GitHub Terms of Service · GitHub Privacy Statement
• WordPress.org Privacy Policy

AtlasAiDev promotions

Your server checks a small JSON file on GitHub Gist (https://gist.githubusercontent.com/) for current AtlasAiDev offers, which are shown as dismissible notices to administrators. No site data is sent.

• GitHub Terms of Service · GitHub Privacy Statement

AtlasAiDev usage data (opt-in only)

Only if an administrator opts in (on the last step of the setup wizard, or in the admin notice), your server sends non-sensitive usage data once a week to AtlasAiDev (https://track.atlasaidev.com/): WordPress, PHP and plugin versions, which modules are enabled, content counts and setup progress. To include the site’s public IP address in this report it calls https://icanhazip.com/. Nothing is sent without opt-in, and you can opt out at any time.

• AtlasAiDev Privacy Policy