Sangar Studio Access Flow is not a generic WCAG accessibility overlay. It is a cognitive adaptation layer designed specifically for people with ADHD, dyslexia, autism, anxiety and sensory overload.
Unlike traditional accessibility plugins that offer static controls (font size sliders, contrast, big cursor), Access Flow combines preconfigured neurodivergent profiles, proactive behavioral adaptation and reading transformations like Bionic Reading and Reading Ruler that the target audience recognizes immediately.
prefers-reduced-motion and the system’s forced-colors mode.gpt-4o-mini) in plain language, short sentences and lists. Hourly cache to save quota.role="dialog" + aria-modal="true" + aria-labelledby + focus trap.aria-pressed on mode buttons, aria-busy during requests.:focus-visible with an amber outline on all controls.prefers-reduced-motion and forced-colors (Windows high contrast).This plugin can connect to OpenAI to generate AI-powered summaries of the current page content. This service is only used when the site administrator provides an OpenAI API key in the plugin settings and the visitor explicitly clicks the “Summarize this page” button in the floating widget. If no API key is configured, or the feature is disabled in settings, no request to OpenAI is ever made.
OpenAI provides the Chat Completions API. The plugin uses the model gpt-4o-mini (configurable via the accessflow_openai_model filter) to generate a plain-language summary of the article the visitor is reading, formatted with short sentences and bullet lists to improve readability for users with cognitive overload, ADHD or dyslexia.
When a visitor clicks the “Summarize this page” button, the plugin extracts up to the first 3000 characters of plain text from the <main> (or <article>, or <body> as fallback) element of the current page and sends it to https://api.openai.com/v1/chat/completions, together with the site’s OpenAI API key in the Authorization header. No personal data, IP address, cookies, user identifiers or browsing history are sent. The text payload is cached server-side (WordPress transient) for one hour, keyed by hash, so identical requests do not re-hit the API.
No data is sent to OpenAI on page load, on visitor identification, in the background or for analytics. The request happens only on explicit user action.
This service is provided by OpenAI, L.L.C.
Site administrators are responsible for disclosing this third-party processing to their own visitors in their site’s privacy policy where required by applicable law (such as the GDPR).