Skip to main content
Bolna Voice AI agents have multilingual support and can have conversations in several languages (see list of all support languages). To ensure natural speech output, it is important to write your prompts in the native script of the target language, rather than phonetically using the English alphabet.

​
Multilingual setup

Bolna supports multilingual configurations. You can add multiple languages to a single agent, set one as primary, and write a separate prompt for each.
Example configurations:
  • English (Primary) + Hindi
  • English (Primary) + Hindi + Dutch
  • English (Primary) + French + Spanish
See the Multilingual Support guide for full setup details, and How Language Switching Works for how the agent decides, mid-call, when to move between them.
Mid-call language switching is handled automatically by Bolna — you don’t need to (and shouldn’t) write switching instructions into your prompt. See How Language Switching Works for the mechanism. The prompt guidance below is about writing each language’s own content well, not about triggering switches.

​
Write the prompt in the native script

Using the correct script:
  • Enables more accurate pronunciation
  • Helps the AI identify the intended language
  • Improves contextual understanding and tone
  • Prevents misclassification as English

​
Tips for writing in native scripts

  • Use Google Input Tools or built-in language keyboards on your phone/laptop.
  • For European languages, make sure to include accented characters (like é, ñ, ü, ¿, ç, etc.).
  • Double-check spellings and punctuation using tools like Google Translate, but avoid relying on it for full sentence correctness.

​
Examples of prompts in native scripts

❌ Incorrect
Bonjour! Comment ca va? Nous allons commencer l’entretien maintenant.
✅ Correct
Bonjour ! Comment ça va ? Nous allons commencer l’entretien maintenant.
Notice the accents (ç, é, ’). These help the AI pronounce words like a native speaker.

​
Common mistake to avoid

Don’t write in “English-style” phonetic spelling for non-English prompts.
❌ Kaise ho?
✅ कैसे हो?
❌ Como estas?
✅ ¿Cómo estás?

​
Best practices for production-grade multilingual prompts

Beyond writing in native script, a few practices consistently separate multilingual agents that stay in the right language from ones that drift:
Don’t leave a language to on-the-fly translation. Write out each response — greetings, confirmations, closings — in every configured language, in its native script, ahead of time.
Add a direct instruction such as: “Speak the variant written for the language the caller is using. Never translate one language’s variant into another, and never read the English text instead.” This is one of the most effective single changes for keeping an agent from drifting into the wrong language mid-call.
Keep common English loanwords (“order”, “OTP”, “support ticket”) in Latin script inside otherwise native-script text, rather than forcing an artificially “pure” translation. This matches how people actually speak and sounds far less stilted.
हिंदी: “आपका order status check करने के लिए मुझे थोड़ा समय दीजिए।”
Politeness and gender are often encoded in verb forms, not word choice — a generic “be polite” instruction doesn’t reliably produce it. Be explicit: specify which verb or address forms to use in each language, e.g. the polite plural (நீங்க / మీరు / ನೀವು) rather than the informal singular in Tamil, Telugu, and Kannada, or consistent feminine verb forms in Hindi for a female persona.
For OTPs, registration numbers, and order IDs, instruct the agent to read them character by character in English regardless of the conversation’s language. This avoids the mispronunciation and misheard digits that translating or reading out a number in another language often causes.
For lines that must not vary — identity confirmation, consent, compliance disclaimers — instruct the agent to speak them exactly as written, without paraphrasing, and pair this with a low temperature so the model doesn’t improvise.
Greetings, hand-ins, and closings that never change shouldn’t be generated by the model at all. In a graph agent, configure them as a static node with a per-language message instead — there’s no room for the model to drift if it never sees the line.
If you only add one thing from this list, add the “never translate” rule. It’s the single change most likely to keep a multilingual agent consistently in the right language.

​
FAQs

You can add multiple languages to a single agent. Set one as primary and add secondary languages as needed. Each language gets its own prompt tab in the Agent Tab.
Bolna agents can dynamically switch between any of the configured languages — this happens automatically, without needing prompt instructions, whenever the caller genuinely moves to another configured language. See How Language Switching Works for exactly what triggers it. If a customer speaks a language not configured on the agent, it will fall back to the primary language.
It depends on your audience and brand tone. Ensure your prompts reflect the appropriate politeness level (e.g., “vous” vs. “tu” in French, or “आप” vs. “तुम” in Hindi) for a consistent and professional experience.
Use Bolna’s Preview Voice feature in your agent’s Languages tab to test generated responses before finalizing your prompts. Adjust words and punctuation if needed for more natural pronunciation.