The Odoo AI chatbot that answers from your real Odoo content

There is a reason the "AI" in Odoo AI chatbot matters more than just a marketing adjective. Odoo websites skew toward two visitor profiles that older rule-based chat tooling handles badly. The first is the B2B technical buyer — someone arriving from a distributor directory, a partner listing, or a LinkedIn referral, already a few questions deep into evaluating whether your product fits a specific engineering context. The second is the international browser — someone whose language may or may not be among the ones your Odoo i18n translations cover, asking about shipping, compatibility, or pre-sales logistics. Both profiles punish scripted bots quickly. The technical buyer phrases questions in ways your intent library never anticipated. The international browser doesn't speak the language your decision tree was authored in. A modern reasoning model, bounded to your own Odoo content, handles both comfortably where rule-based flows stall.

Grounded generation is the feature that makes the AI side of this workable in production. A raw language model pulled from a vendor's API will happily invent plausible-sounding product specs, delivery promises, or refund policies — which on an Odoo eCommerce site translates to commercial exposure within minutes. Asyntai's approach pins the reasoning layer to a retrieval index built from your Odoo site plus whatever private documents you upload. When a visitor asks something, the model drafts a reply only after consulting the indexed material; if nothing relevant turns up, the reply acknowledges the gap and offers a handoff rather than fabricating. That guardrail is the difference between AI you can leave running unattended on a live Odoo site and AI that needs a human editor watching every transcript.

Rolling out that AI capability onto an Odoo stack is where the module matters. Odoo administrators and integrators live inside a culture of addons — installable via the Apps screen, configurable on the standard res.config.settings surface, cleanly removable when a project wraps up. Asking an Odoo team to paste a random JavaScript tag into a theme header file feels structurally wrong in that environment, even when it would technically work. The asyntai_chatbot module fits the expected pattern. It appears in Update Apps List, installs with a click, adds a dedicated app to the Odoo launcher, exposes settings inside the familiar admin UI, and removes itself cleanly on uninstall. For the subset of Odoo tenants where custom modules aren't allowed — primarily Odoo Online — we ship the same AI under a JS snippet delivery, so no Odoo hosting choice locks a team out of the product.

The version matrix deserves attention because Odoo's internal APIs do drift between major releases. A module that works against Odoo 17.0's website controllers won't necessarily load under 19.0's updated routing, and the reverse is also true. Rather than shipping one "universal" build that probably breaks somewhere, the Asyntai archive carries a separate, tested module per supported release: 16.0, 17.0, 18.0, and 19.0. The installation instructions tell you to pick the folder matching your major version — a 60-second choice that saves the debugging work of chasing API drift errors later. Teams running heavily customized Odoo forks or unusual minor versions that aren't in the matrix can still use the JS snippet path; the chatbot's answer quality is identical regardless of how the widget gets onto the page.

Training the AI on your content happens automatically on first connect. Once the OAuth popup closes, Asyntai fires off a crawl across every reachable page on your Odoo Website — home, About, services, product pages, category trees, blog posts, contact, legal, terms, every public URL the Website module serves. The crawler converts each page into an embedded representation and stores those in the retrieval index that will feed answers from that moment on. Re-crawls run on a cadence so that adding a new product or editing a policy flows into the chatbot's knowledge without any manual "retrain" step. For the 20–40 per cent of your content that isn't public — partner-specific pricing sheets, manufacturing tolerances, support SLAs that only exist in signed contracts — the dashboard accepts PDF uploads and pasted text, and the same retrieval pipeline indexes that private material the same way.

Custom instructions sit alongside the retrieval layer and shape how the AI actually replies. These aren't prompt-engineering incantations; they're plain sentences. "For enterprise inquiries above 500 units, always invite the visitor to request a custom quote." "Never commit to same-day delivery — always frame it as 'typically'." "If a visitor mentions regulated industries, flag that the technical team will verify compliance before shipping." The AI weighs each instruction every time it drafts a reply, so behavioural tone, commercial hedging, and escalation rules stay consistent across every conversation without you rewriting a script tree. It is the closest equivalent to briefing a new junior salesperson — you tell them how the company talks, and they hold that line.

Multilingual coverage is where the AI angle pays off most visibly on international Odoo sites. Odoo's own i18n tooling lets you translate site content into a chosen set of languages, but that set is always bounded — translation work costs money, so you stop at your priority markets. An AI chatbot built on a multilingual foundation has no such bound. Asyntai carries 36 interface locales and the reasoning layer picks up a visitor's language off their opening line, producing native-language replies regardless of whether your Odoo pages have been localised into that tongue. A Greek visitor to a German-only Odoo site still gets Greek answers drawn from your German content. That asymmetry — your content in a few languages, the AI replying in many — is what turns Odoo sites serving 10 or 15 markets into something that actually feels localized at the conversation level.

Personalization for logged-in Odoo portal users runs through the User Context feature. Asyntai's widget exposes a window.Asyntai.userContext object; the Odoo theme fills it in before the widget initialises, with whatever portal data you choose to surface. Name, company, plan tier, last invoice status, active support contract, region — you decide. The AI uses that context when drafting replies, so an enterprise customer asking about support response times gets an answer pegged to their actual SLA tier rather than the generic public copy. None of this requires wiring up a permissioned API from Odoo to Asyntai — you control what goes into the context object, and nothing else is readable. On Standard and Pro plans, this account-aware behaviour is on by default; on lower tiers, the chatbot operates in anonymous mode.

When an AI chatbot can't close out a conversation on its own — a scoping discussion too complex for the indexed material, a genuine enterprise pricing negotiation, a regulated industry question requiring human judgement — the lead capture flow kicks in. The AI asks for email and relevant business context, assembles the full transcript, and drops the lead into your Asyntai dashboard. Optional email notifications push the transcript to a sales inbox in real time. Odoo-native teams typically pipe that notification email into an Odoo CRM inbound alias, which creates a lead record automatically; the AI effectively becomes the top-of-funnel that populates your Odoo pipeline without a human watching the chat queue.

Multi-site Odoo is where the AI scaling story lands. One Odoo instance often serves multiple brands, regional subdomains, or client deployments under a multi-company configuration. Asyntai attaches independently to each site — separate knowledge bases, separate instructions, separate widget branding — with plan tiers defining how many sites one account can handle: free tier at 1, Starter at 2, Standard at 3, Pro at up to 10. An agency running Odoo for a portfolio of clients can centralize AI chat delivery on a single Pro account and still keep each client's chatbot fully siloed in terms of training data and tone.

Intelligence, grounded to your own Odoo content, delivered through a native module where the Odoo platform allows it and through a JS snippet where it doesn't — that is the shape of the Asyntai Odoo AI chatbot. The AI is what makes it worth installing; the Odoo module is what makes it sit comfortably inside the platform your team already administers. Install from addons, connect in a click, let the crawler learn your site, upload anything private, write a handful of custom instructions, and the chatbot starts carrying the visitor-facing layer on every page of your Odoo Website — in every language your visitors happen to arrive in.