A Moodle AI plugin that installs like any local plugin

Institutions that run Moodle tend to have a very specific reflex when evaluating any new add-on. The question is rarely "is the feature good" first — it is "does this fit how we install things, and will an upgrade break it." A Moodle AI plugin that requires editing a theme's layout file, or that asks IT to sprinkle JavaScript into config.php or a boost layout, already fails that test. The Asyntai Moodle AI plugin is built with that reflex in mind: it arrives as a local plugin component, which is the category of Moodle plugin with the smallest possible blast radius. It does not register a gradebook type, it does not add an activity module, it does not touch course formats. It installs, it exposes a settings page, and it renders an assistant on top of whatever theme the institution uses.

The mechanical install takes a handful of minutes. An admin logs into Moodle, goes to Site administration → Plugins → Install plugins, drops the local_asyntai.zip file into the uploader, and lets Moodle run its validation and database-upgrade cycle. The plugin is compatible with Moodle 3.9 upwards, which covers both the older LTS releases some universities stay on for stability and the current 4.x line others have migrated to. Because it is a local plugin, the upgrade cycle is quick — a minimal schema gets created for storing the Asyntai site pairing information, and control returns to the admin. From there, Site administration → Plugins → Local plugins → Asyntai opens the plugin's settings page, where a single Get started button begins the pairing process with an Asyntai account.

Pairing is the part that typically worries IT more than the install itself. On a lot of tools, it means logging into a vendor portal, copying an API key, pasting it into Moodle, saving, praying the value was not truncated, and then emailing support when nothing works. The Moodle AI plugin avoids that sequence entirely. Clicking Get started opens a browser tab that handles authentication against the admin's Asyntai account, confirms the Moodle site identifier, and closes on its own. The pairing is stored server-side between the two systems rather than as a user-pasted secret, which removes both a support-ticket source and a small attack surface. After that single confirmation, the assistant is live across every Moodle page — login, dashboard, course listings, individual course and activity pages — with no further configuration required.

What the assistant actually does in a Moodle context is where the practical return shows up. Students arriving at a learning management system do not arrive with fully formed questions; they arrive with half-formed confusion. "Where do I submit this." "Is this course still open." "How do I reset my enrollment key." "When is the midterm." Historically, the answer to all of those has been a combination of a course page they overlooked, a help document nobody reads, and an email to the program office. The Moodle AI plugin puts a conversational layer in front of all of it. The assistant is trained on the public pages of the site — course catalog, category pages, any outward-facing description — and then enriched with documents the admin uploads into the Asyntai dashboard: student handbooks, academic calendars, IT help articles, refund policies, code-of-conduct documents, and anything else that normally lives in a PDF buried on the Help page.

Instructional designers tend to be the power users of the configuration side. The Asyntai dashboard exposes a plain-English instructions field, which is where institution-specific behaviors live. A designer can write rules like "Always point students asking about accessibility to the Disability Services office page." "For questions about graduate-level courses, mention that prerequisites exist and link to the prerequisite checker." "If a question mentions grade appeals, refuse to guess and route the student to the registrar." Those rules are picked up instantly — no retraining, no custom model, no engineering ticket. The designer writes the rule today, and tonight's students see the new behavior.

For course coordinators, the interesting part is the conversation log. Every exchange between a student and the assistant is stored in the Asyntai dashboard, searchable by keyword, date, and topic. After a couple of weeks of traffic, patterns surface: the single most-asked question of the term, the course whose page generates the most confusion, the week where pre-enrollment questions spike. That list is editorial ammunition. If three hundred students asked the same question about a hybrid attendance policy this semester, the course's description page needs a rewrite. If enrollment keys trigger a weekly flood of identical questions, the enrollment instructions page needs one paragraph added. The Moodle AI plugin answers the questions now; it also tells the coordinator which pages to fix so fewer students need to ask at all.

Passing logged-in student data into the assistant is the feature that most distinguishes a Moodle-native plugin from a generic chat widget. Standard and Pro plans ship a window.Asyntai.userContext entry point the Moodle site can populate during page render. The institution decides what goes into it — typically a first name, an enrolled course list, a program or cohort code, maybe a progress flag — and the assistant uses the context to personalize its replies. A logged-in student asking "when is my next assignment due" can receive an answer grounded in the data the institution chose to pass, without the assistant needing to call Moodle's web services directly. Institutions that do not want to share any student data simply leave the entry unset; the assistant then treats every visitor as anonymous.

Language coverage is another feature Moodle admins flag early, especially at institutions with international recruitment. A Moodle site configured with a handful of Moodle language packs is one layer; a chat assistant that still only replies in English is a separate problem. Asyntai's assistant UI renders in 36 languages, and the language of each reply is detected from the student's own message at conversation time. An Arabic-typing student receives Arabic, a Vietnamese-typing student receives Vietnamese, a French-typing student receives French, regardless of the language the Moodle interface has been set to. For programs serving mixed cohorts or branch campuses abroad, this is the feature that makes the plugin usable at all.

Security and privacy reviews usually happen before an institution is allowed to approve any outside tool, and the Moodle AI plugin is designed to pass them without surprises. The assistant runs client-side as a widget; it never reaches into the Moodle database, it never accesses the gradebook, it never pulls user profiles on its own. Anything the assistant knows about a user is data the institution chose to push through the user-context object. Uploaded training documents sit in the Asyntai dashboard under the institution's account. Conversation logs are retained for analytics; retention and export options are exposed in the dashboard. Documentation covering the data flow is available on request during a procurement review.

Pricing maps cleanly to real institutional usage. The free tier — 100 messages per month — is calibrated for a pilot on a single course, a staff sandbox, or a department-sized test before a broader rollout. The paid entry plan at $39 per month covers 2,500 messages, comfortable for a small faculty or a continuing-education program. Higher tiers scale to department-wide and university-wide rollouts, with the multi-site allowance (one, two, three, and up to ten) covering institutions that run separate Moodle instances for undergraduate, graduate, continuing-education, or international divisions. Billing lives in a single Asyntai account, which often matches the institutional procurement model better than per-instance licensing.

Adoption in higher education is a category that genuinely pays. A meaningful share of Asyntai's paying customers are education organizations running Moodle, and the reasons they cite are consistent across institutions: the helpdesk queue during add/drop week, the repetitive tier-one questions that academic advisors spend their mornings answering, the language-coverage gap for international students, the flood of "how do I" questions any new LMS rollout generates for the first semester. The Moodle AI plugin slots into each of those without asking any existing staff member to change their workflow. Advisors keep doing advising, but without the twenty daily emails asking where the course catalog is. IT keeps doing IT, but without the password-reset and enrollment-key tickets that a well-trained assistant deflects. Instructors keep teaching, but without having to answer the same syllabus questions in every course's discussion forum.

Removal, which is always worth thinking about up front, is equally boring. Uninstalling the Moodle AI plugin is the standard Moodle local-plugin uninstall: visit Site administration → Plugins → Plugins overview, find Asyntai under Local plugins, and choose Uninstall. Moodle removes the plugin's database tables and files, the assistant stops appearing, and that is the end of it. Your Asyntai dashboard still contains the conversation history and any training documents uploaded — you can export or delete them from the dashboard directly, and the Asyntai account itself can be closed with a single request if the institution has fully decided to move on.

For any Moodle instance where the student support queue is already strained, or where international enrollment is growing faster than staffing, or where tier-one helpdesk volume dominates the workday, the Moodle AI plugin is a disproportionately small install for a disproportionately large operational change. The plugin goes in through the normal Moodle flow, the pairing takes one click, the training takes a few uploads, and by the end of the same afternoon the LMS has a 24-hour multilingual assistant that knows the institution's own documentation. It does not require Moodle plugin development, it does not require engineering hours, and it does not require committing to a new vendor ecosystem — just a local plugin, a pairing click, and a dashboard login.