PRO
    Phone Systems
    Advanced
    Ready

    Bilingual AI Receptionist (English + Secondary Language)

    AI answers calls in English OR Spanish (or Mandarin/Vietnamese/Tagalog/Portuguese), detects language in 3 seconds, switches scripts, books appointments. Captures the 20-40% of leads your competitors lose to language barriers.

    Build it yourself, free

    This blueprint is free. Copy the master prompt below, build it with your own time, and the only outlay is a few low-cost tool subscriptions. Use it inside your own business, or sell it to clients. The numbers further down are what you can charge, not what you pay.

    The Bilingual AI Receptionist Blueprint
    Read the story behind this blueprint

    The Bilingual AI Receptionist Blueprint

    Read the article

    Who you sell it to

    Service businesses in metros with significant non-English-speaking populations. Hard targets: HVAC/plumbing/roofing/landscape/cleaning/auto repair in Phoenix, LA, Houston, San Antonio, Dallas, Miami, NYC, Chicago, San Jose, Newark, Orange County. Also high-value for: dental offices, med spas, family restaurants, immigrant-owned retail in these markets.

    The pitch lands hardest with owners who can name a recent Spanish-speaking lead they lost because nobody at the office could communicate with them, or who have a Spanish-speaking employee they pay $50k/year specifically to handle phone calls. Particularly powerful for businesses with an English-speaking owner/operator who's been told repeatedly "you really need to hire someone bilingual" but hasn't because of cost or time.

    If you sell it, here is the money picture:

    The first tile is the value of your client's own jobs, which is the reason they will pay you. The other two tiles are what you charge them.

    What your client's own work is worth
    Captures additional $50-200k/yr in lost-to-language-barrier leads
    Their revenue, not your fee
    Setup fee you charge
    $2,500
    One-off, paid to you
    Monthly retainer you charge
    $697/mo
    Recurring, paid to you

    How it works

    A customer calls the business. Within the first 3-5 seconds of the conversation, the AI detects whether they're speaking English or the configured secondary language (Spanish by default, operator can configure Mandarin, Vietnamese, Tagalog, Portuguese, French, Russian, Korean, or Arabic during setup). The AI then conducts the entire conversation in that language using culturally appropriate phrasing, not robotic translation but natural native-feeling speech.

    The AI captures the caller's name, issue, address, urgency, and books an appointment OR routes to the appropriate team member. CRITICAL ROUTING LOGIC: - If caller is English-speaking AND English-speaking tech available → standard handoff - If caller is Spanish-speaking AND Spanish-speaking tech available → route to that tech - If caller is Spanish-speaking AND no Spanish-speaking tech available → AI completes the booking entirely OR makes specific callback promise: "Tendremos a Carlos llamándole en los próximos 30 minutos" - Failed callback promises = lost customer + bad word-of-mouth in tight-knit immigrant communities. Tracking + alerting on missed callbacks is required.

    The owner gets call logs with language detected, conversation transcript in original language, AND English translation for review. Analytics dashboard shows conversion rate by language, missed-call rate by language, and dollar value per language. For an HVAC contractor in Phoenix where 35% of incoming calls are in Spanish and they're currently converting only 12% of those (vs 28% English), this tool typically lifts Spanish conversion to 24-26%, directly recovering $80-150k/year in revenue previously lost to language barriers.

    What to expect

    This is the most complex tool in the library, bilingual voice AI with real-time language detection, dual-language scripts, language-aware routing, bilingual analytics. Expect Base44 to take 30-45 minutes to build. Likely follow-up prompts:

    If language detection isn't working in first 5 seconds: "Language detection is taking too long or misclassifying. ElevenLabs Conversational AI should handle this natively with multilingual_v2 model. Verify: (1) agent configured with multilingual_v2, (2) both English and Spanish voice IDs configured, (3) initial greeting plays both languages so caller's response triggers natural code-detection, (4) LLM receives the audio language signal from ElevenLabs as part of turn context."

    If AI's Spanish sounds like Google Translate: "Spanish scripts are sounding like Google Translate, not native Spanish. LLM system prompt for Spanish calls must explicitly: (1) instruct culturally native phrasing (not literal translations), (2) use 'usted' formality by default, (3) include natural Mexican Spanish idioms and warmth markers ('con gusto', 'no hay problema', 'a la orden'), (4) reference voice samples from native Mexican Spanish speakers if provided. If quality remains robotic, source 5-10 sample transcripts of actual Spanish-speaking phone receptionists from target region and feed as few-shot examples."

    If callback queue isn't escalating on missed deadlines: "Callback promises past deadline aren't escalating. Scheduled function checkCallbackDeadlines must run every 5 minutes (not 15, too slow), check all CallbackPromise records where status='pending' AND promise_deadline < now, fire URGENT SMS to owner AND assigned tech. Dashboard must show these in red with alarm sound on mobile."

    If ElevenLabs voice quality is uneven between languages: "Voice quality inconsistent, English voice sounds great, Spanish sounds artificial. ElevenLabs offers different voices with different multilingual quality. Some voices are optimized for one language.

    Solution: configure two separate ElevenLabs Agents, one with voice optimized for English (Adam, Rachel), one with voice optimized for Spanish (Antoni, Bella in Spanish mode, or Spanish-specific voice). System switches between agents based on detected language."

    If Twilio call recording disclosure isn't being added: "Some states require all-party consent for call recording (CA, FL, IL, others). Add bilingual disclosure at call start: 'This call may be recorded for quality purposes. / Esta llamada puede ser grabada para calidad.' Either as pre-greeting Say block or integrated into bilingual greeting."

    If A2P 10DLC is rejecting Spanish SMS campaigns: "Spanish A2P campaigns sometimes face additional carrier scrutiny. Provide more detailed use case: 'Transactional appointment confirmations sent in Spanish to our Hispanic customers who initiated contact with our business in Spanish. These are bilingual customer care communications, not marketing.' Provide 5 sample Spanish messages.

    May need to register as separate campaign from English."

    THE BIGGEST GOTCHA: ElevenLabs voice quality calibration. Same voice can sound great in one language and robotic in another. Test thoroughly with native speakers of the secondary language BEFORE going live.

    A bad-sounding Spanish AI does more damage than no Spanish AI at all.

    SECOND BIGGEST GOTCHA: The callback queue. If AI promises a Spanish-speaking tech will call back within 30 minutes and that doesn't happen, customer is lost permanently AND tells the community. Escalation logic must be aggressive, promise broken = owner gets immediate SMS, dashboard goes red, followup happens NO MATTER WHAT.

    Master prompt

    You are an expert full-stack builder. Build a production-ready tool that… (the complete master prompt, custom input fields, build phases, pricing guidance and sales script are part of this blueprint.)

    Create a free account to unlock the full blueprint

    Master prompt, editable client fields, phase-by-phase build steps, pricing and the sales pitch.

    Build phases

    Step-by-step build phases unlock when you log in.

    Required tools

    The AI builder

    App builder hosting database, language-aware routing logic, admin dashboard, callback queue management

    Builder plan $50/mo

    Sign up

    ElevenLabs Conversational AI

    Multilingual voice AI, handles real-time conversation in English + secondary language with natural voice quality

    $22-330/mo depending on call volume (Starter for testing, Business for production)

    Sign up

    OpenAI (GPT-4o) or Anthropic Claude

    LLM for intent classification, urgency detection, transcript translation to English for owner review

    ~$0.03 per call (~$20-50/month for typical business)

    Sign up

    Twilio

    Phone number, voice webhook to ElevenLabs streaming, SMS confirmations, call recording

    $1.15/mo per number + ~$0.013/min voice + ~$0.008 per SMS

    Sign up

    Native speaker for voice quality verification

    One-time verification that AI's secondary language sounds natural before launch

    $25-50 one-time via Fiverr or community contact

    Sign up

    AppBuild.DIY (Vibes)

    Wraps the Base44 owner/admin dashboard into native iOS/Android apps plus a polished PWA, branded to the business. The main relevant Native Capabilities plugin here is push notifications: when a callback promise is closing in on its 15/30/120-minute deadline, a lock-screen push hits the owner and assigned tech directly, backing up the Twilio SMS escalation in case a text gets missed or muted. Handles branding and app store submission.

    Free up to 10 users, $49/mo for 1,000 users, $99/mo unlimited

    Sign up

    Pricing

    Pricing guidance unlocks when you log in.

    Sales pitch

    The outreach script unlocks when you log in.