Joomla chatbot extension you upload straight through Extension Manager

Joomla occupies a specific corner of the CMS market. It's been around long enough that the administrators who keep using it tend to be opinionated about how extensions ought to behave, and the communities around Joomla are unusually international — Italian agencies, German universities, Spanish-speaking membership sites, Polish publishers, Brazilian news portals. Two things follow from that. First, anything that drops onto a Joomla site has to look and feel like a native Joomla extension, not a third-party hack wedged into the template. Second, whatever you ship has to speak the visitor's language on day one, because a non-trivial slice of Joomla traffic is multilingual by design. The Asyntai Joomla chatbot was built with both constraints in mind — a proper plg_system_asyntai system plugin that installs through the standard Extension Manager, configured through the plugin edit screen every Joomla admin already knows, combined with a 36-language AI layer that handles visitor languages the Joomla site itself may not have a language pack for.

Installation is deliberately unexciting. You download the archive from the Asyntai dashboard — a single zip that covers both Joomla 4 and Joomla 5 through the unified CMSPlugin API. In the Joomla admin you open System → Install → Extensions, drop the zip onto the upload area, and the extension is registered. Next you pop over to System → Manage → Plugins, filter for "system," find "System - Asyntai - AI chatbot" and flip the enable toggle. That's the Joomla side done. No template override, no template.php edit, no manual script tag in the head of your preferred template, no separate component to install. Joomla admins who have done this dance for a hundred other extensions will do it here without reading documentation.

The connect step is where the plugin departs from the older pattern of pasting an API key into a field. Open the Asyntai plugin to edit, and the plugin's own admin injection adds a Get Started button to the edit screen. Click it, a popup window opens pointing at asyntai.com, and the normal Asyntai sign-in or sign-up flow runs. The popup closes once the account is linked, a lightweight polling script hands the site ID back to the Joomla plugin via an authenticated AJAX endpoint, and the plugin writes site_id into its params through com_ajax. From that point the site is connected. If the popup is blocked, a fallback link opens the connect window in a new tab — no admin gets stuck. Resetting the connection later is a single link inside the same plugin edit screen.

Content training starts immediately after connect. The Asyntai crawler walks the Joomla frontend — every article, every category blog view, every menu item, every third-party component route that exposes readable HTML at an SEF URL. Most Joomla sites aren't purely com_content anymore; they lean on K2 for publishing, VirtueMart or HikaShop for ecommerce, Fabrik for data-heavy pages, EasyBlog or JomSocial for community output. None of that matters to the crawler because it operates on the rendered HTML the visitor sees, so a chatbot on a Joomla/VirtueMart shop reads the same pages a human customer would read before asking about shipping, returns or product variants. That crawled layer becomes the base of the bot's answer repertoire.

Private material lives separately. Joomla's ACL restricts plenty of useful documentation to Registered, Special or Super User access groups — knowledge base articles for paying subscribers, rate cards for partner resellers, internal onboarding manuals for new editorial staff. Exposing those to a crawler would defeat the whole point of ACL, so the Asyntai approach is the inverse: upload the documents directly through the dashboard as PDFs or pasted text, and the chatbot references them without ever needing to pull them from Joomla. The bot's knowledge is therefore the union of what's public on the Joomla frontend and what's been explicitly given to it via upload — a cleaner separation than trying to hand out authenticated crawler tokens.

Multilingual Joomla deployments benefit disproportionately. Joomla's native multilingual mode — distinct language codes per article, language associations, the sitewide Language Switcher module — is arguably the most mature multilingual story of any open-source CMS. Editors build a Swedish version of an article, associate it with the English one, and Joomla routes visitors based on their language preference. The limit there is that editorial translation only covers the languages the team has the capacity to write in. A Swedish publisher running Joomla might offer editorial content in Swedish and English but still see traffic from Norwegian, Danish, Finnish and German readers. The Asyntai chatbot detects each visitor's language from their first message and replies in that language, so the conversation happens in the visitor's native tongue even when no editorial pack exists for it. The Joomla site's language switcher keeps doing what it always did; the chatbot fills the long tail.

Agency teams running multiple Joomla sites for multiple clients are a second population the platform serves well. A single Asyntai account can own several connected sites — one per client domain — with the plan tier determining how many: the Free tier hosts one site, Starter two, Standard three, Pro up to ten. Each site gets an independent training corpus and its own personality. A travel agency client's Joomla chatbot can be upbeat and playful; a law firm client's Joomla chatbot can be formal and hedged; a medical-equipment supplier's Joomla chatbot can stick strictly to spec-sheet answers. All three are configured from the same Asyntai dashboard, saving the agency from juggling three separate subscriptions while still keeping each client's voice distinct.

User Context is the mechanism that lets the chatbot go beyond anonymous visitor mode and behave like a member-aware assistant for logged-in Joomla users. On Standard and Pro plans, you expose a window.Asyntai.userContext object before the widget loads — easiest through a small custom module in the debug position that reads the current Joomla user's name, group, and any relevant subscription data (Akeeba Subscriptions tier, PayPlans level, DocMan access, whatever applies). The chatbot incorporates that context into each response: a logged-in Registered visitor gets an answer different from an anonymous one; a paying Pro member gets an answer different from a free member. No data leaves Joomla's server through a permissioned API — you choose exactly what's put into the JavaScript object, so the user's actual password hash, payment tokens, or protected fields never cross into the chat layer.

Caching is usually the next administrator concern, because Joomla sites that get real traffic live or die by their caching stack — System - Cache plugin, JotCache, LiteSpeed Cache for Joomla, CDN caching via Cloudflare. The Asyntai plugin injects its widget loader at onAfterRender, which runs after page composition, so whatever caching layer captures the output sees a complete HTML document with the script tag already in place. That cached copy serves correctly from the CDN or disk cache without needing to be busted every time you change something in the Asyntai dashboard, because the dashboard settings take effect client-side via the widget's own config fetch — the cached HTML remains valid. Administrators running aggressive Joomla caching setups don't need to add the Asyntai script to any exclusion list.

Conversation analytics give Joomla administrators something Joomla itself doesn't natively offer. Every chat thread lands in the Asyntai dashboard grouped by topic, frequency and visitor language. A Joomla-based university portal can see which admissions questions dominate; a Joomla-based Moodle marketing site can see which course enquiries recur; a Joomla-based regional news publisher can see which archive topics drive curiosity. Those patterns feed back into content strategy — new articles to write, existing articles to clarify, FAQ pages to update — and the chatbot measures its own contribution by tracking how often conversations end in a captured lead or a useful answer versus an unresolved question.

Lead capture is the commercial hinge for Joomla sites that treat their site as a business asset rather than a pure publishing outlet. Agencies selling Joomla development time, consultancies using Joomla for client portals, SaaS-style products running their marketing site on Joomla — all need reliable lead routing. When the chatbot can't resolve a thread, it asks for an email, records the entire conversation, and files the record into the Asyntai dashboard. Enabling email forwarding pushes the lead into whatever inbox your team monitors, and from there it routes into Joomla's com_contact inbox, a CRM integration, or a simple Slack channel. Because the full transcript travels with the lead, the follow-up human already knows what the visitor tried to do, which removes the dead "what were you asking about?" opening most cold lead responses suffer from.

Cost-wise, the Joomla chatbot sits at a level Joomla-scale businesses can absorb without ceremony. A free tier handles up to 100 messages so you can run the plugin on a staging build or a small club site without committing. The $39 per month Starter plan unlocks 2,500 messages, which comfortably covers most single-site Joomla properties — a municipality portal, an association site, a regional news platform, a small online shop. Higher plans push the message ceiling up and enable User Context for logged-in personalisation. For Joomla teams accustomed to paying annually for a template club and a component bundle, adding a chatbot subscription at that tier sits inside the existing expectations for their stack rather than becoming a new line item requiring board approval.

Rolling the Joomla chatbot out follows a tidy path. Upload the zip through the Extension Manager, enable the system plugin, connect your Asyntai account through the popup, let the crawler index your Joomla frontend, upload any ACL-protected documents you want the bot to reference, write a short set of plain-sentence instructions that capture how you want the voice to sound, open the site in an incognito window to sanity-check real visitor interactions, and ship it. Nothing in the Joomla stack changes — articles still publish through com_content, menus still route through com_menus, templates still render through your preferred framework, caching still caches, SEF still rewrites URLs. The chatbot quietly sits on top of all of that, answering the visitor questions your Joomla frontend would otherwise have to handle via contact forms, email replies, or lost traffic.