Yes. Any language the underlying LLM supports. OpenAI, Anthropic, and Google models cover 50+ languages including all major European, East Asian, and Arabic varieties, plus less common ones at decreasing fluency. For high-volume languages (Spanish, French, German, Mandarin, Portuguese, Italian, Japanese, Korean) responses match native quality. For others, test the bot's output against a teacher's read before launching.

Yes. Configure your placement questions in the system prompt and the bot scores responses to recommend a level. The placement is recorded with each answer so a teacher can audit it on the first day of trial. Most schools tune the prompt over the first month, watching where the bot over-places (often confident-sounding B1s who are really A2) and tightening the diagnostic accordingly.

If DELE, DELF, TestDaF, JLPT, Goethe-Zertifikat, CPE, or any other exam prep is on a WordPress page, the bot reads the schedule, frequency, and intensity. It also matches the exam date to the right cohort, so a prospect with DELE B2 in November gets routed to the September intensive rather than the spring evening course.

Yes. Multibot lets you run a Spanish bot, a French bot, and a German bot, each on their respective course pages with their own tone, source pages, and exam-prep specialisation. Conversation logs stay separate per bot, which makes it easier to review enrolment patterns by language without filtering. Most multi-language schools run three to five bots.

Yes. Hand off to Acuity, Calendly, or any LMS with a public booking URL via deep link or webhook. The bot collects preferred slot, level, and language in conversation, then either passes the prospect to your booking link with the slot pre-selected or fires a webhook that creates the appointment server-side. Most schools use deep links.

It's not a substitute for a teacher's placement, but for self-placement at A1 to B1 it matches teacher-placed students more than 80% of the time when prompts are tuned. Above B2, accuracy drops because the diagnostic gets harder to automate. The trial class confirms the level in either case, and the placement log lets you measure drift and improve the prompt over the first quarter.

Yes, and that's the right pattern for B1 and above. Configure the prompt to ask the diagnostic questions in the target language and rate the prospect's responses on grammar, vocabulary, and tone. The bot then recommends a level based on how the prospect handles the language, not how confidently they describe their own ability, which is where most self-placements break down.

The bot handles code-switching natively because the underlying models do. A heritage Spanish speaker who writes English with embedded Spanish gets a response in the language they led with, and the bot doesn't penalise the mix in placement. Many schools run a heritage-track placement question in the prompt to route those prospects to the right course rather than a generic A2 or B1.