--- title: "Voice Bot for Auto Shops That Books & Charges on the Call" description: "Build a voice bot for your auto shop that schedules appointments, checks live bay availability, and collects deposits—all in one call. Full setup in 12 weeks." slug: "bot-de-voz-para-taller-mecanico-que-agende-servicios-en" url: "https://catalizadora.ai/blog/bot-de-voz-para-taller-mecanico-que-agende-servicios-en" cluster: "implementacion-ia/bot-voz-taller" author: "Pablo Estrada" published_at: "2026-05-13T19:24:47.564561+00:00" updated_at: "2026-05-13T19:25:31.598525+00:00" read_minutes: "7" lang: "en" --- # Voice Bot for Auto Shops That Books & Charges on the Call > Build a voice bot for your auto shop that schedules appointments, checks live bay availability, and collects deposits—all in one call. Full setup in 12 weeks. A voice bot for an auto shop that books services works when five things come together: natural Latin Spanish voice, total latency under 800 ms, live bay slot integration, deposit collection within the same call, and clean escalation to a human when needed. In 2026, the complete system ships in 12 weeks with MAGIA / Forge at $20,000 USD for chains with three or more locations. What used to take 30 engineers and 18 months now ships in weeks. The operational problem at a shop in Bogotá or Medellín is brutally concrete: the phone doesn't stop ringing between 7 and 9 a.m., exactly when mechanics are pulling into bays. Every missed call is a customer walking across the street to the competitor. A well-built voice bot answers all 50 morning calls in parallel—no one waits. ## What the bot does on an auto shop call, exactly Seven operational steps in under 3 minutes per call. - **Greeting in the shop's voice**: "Taller Acelera, good morning, how can I help you?" - **Intent detection**: quote request, scheduling, vehicle status, complaint, other - **Vehicle data capture**: make, model, year, plate, mileage - **Initial diagnosis**: service type (oil change, brakes, alignment, general mechanics) - **Live bay slot lookup**: offers three real time slots with assigned bay and mechanic - **Confirmation with deposit**: payment link via SMS or WhatsApp, appointment locked on receipt of confirmation - **Reminder**: automated message sent 24 hours and 2 hours before the appointment If the customer is on the fence, the bot offers a WhatsApp summary with a detailed quote and self-service scheduling. ## Tech stack for a shop running 12 hours a day | Layer | Technology | Why | |---|---|---| | Inbound telephony | Twilio Programmable Voice with local Bogotá number | Colombia coverage | | ASR | Deepgram Nova-2 Latin Spanish | 150 ms latency | | LLM | Claude Sonnet with RAG from the shop's service catalog | Streaming under 250 ms | | TTS | ElevenLabs Turbo v2.5 with owner's cloned voice | Authentic voice | | Orchestrator | Python with WebSocket and Redis | Handles barge-in | | Database | Supabase PostgreSQL with RLS | Live bay slots | | Calendar | Google Calendar API or Calendly | Bidirectional sync | | Payments | Wompi, Mercado Pago, or ePayco payment link | Colombia coverage | | Logs | Append-only with timestamp and recording | Fully auditable | ## Real case: conversational bot with 26.5% conversion A sewing school in Mexico City implemented a 7-phase conversational bot (discovery, informing, proposing, booked, lost) with HubSpot integrated. Although the final channel was written WhatsApp rather than phone voice, the conversational architecture with slot guardrails and close logic applies identically to an auto shop. - **113 conversations** managed in five months by the bot - **30 confirmed bookings**, 26.5% bot-to-appointment conversion - **79 automated follow-ups** at 24, 72, and 168 hours - **57 handoffs** to a human at the closing moment - **5 enrollments** closed from the funnel - **1.36 million pesos** closed - **32.9% organic conversion** versus 14.1% from paid ads For a shop handling 50 daily calls with similar conversion, the bot would automatically schedule 13 services per day—equivalent to 250 additional services per month without adding headcount. ## How the bot checks live bay availability This is the piece that separates an operational bot from a greeting bot. The shop's system must have digitized slots (Google Calendar, Calendly, or proprietary software). The bot queries via API in real time during the call. Bot prompt: "Does tomorrow, Thursday, work for you? I have two options: 8 a.m. with Carlos or 2 p.m. with Andrés." If the customer confirms, the bot reserves the slot via API before hanging up and sends confirmation via WhatsApp. If the shop is still running on Excel or a paper calendar, the first step is to digitize: a basic bay management system is built on Supabase in one week, included in MAGIA / Forge. ## Deposit collection integration Three relevant payment gateways in Colombia: Wompi (Bancolombia), Mercado Pago, and ePayco. The bot generates a payment link with the deposit amount (typically 20–30% of the quoted service) and sends it via SMS or WhatsApp mid-call. When the system detects a confirmed payment (webhook), the bot confirms the appointment by voice: "Perfect, I received your payment—your appointment is confirmed for tomorrow at 8 a.m." Without a deposit, the show-up rate drops from 90% to 50–60%. With a 20% deposit, it climbs to 95%. That difference pays for the system within a quarter. ## How to prevent the bot from double-booking an occupied bay With an optimistic lock on the slot during the conversation. When the bot offers three time slots, it marks all three as "temporarily reserved for 5 minutes" in the database. If the customer selects one, the other two are released immediately. If the call drops without confirmation, all three are released after 5 minutes. This prevents double-booking even when two bots are running simultaneous conversations. ## Latency: the detail that determines success Three specific optimizations that push total latency under 800 ms. One: ElevenLabs streaming—don't wait for the full audio to generate, start playing while it renders. Saves 300 ms. Two: ASR with `utterance_end_ms` set to 500 ms, which detects end of turn before absolute silence. Three: LLM with streaming—first token fires the first word of TTS. Without these three techniques, the bot sounds robotic. With them, customers end calls asking, "Who did I just speak with? Who booked my appointment?" ## Next steps For shops with three or more locations in Colombia handling more than 1,000 calls per month, [MAGIA / Forge](https://catalizadora.ai/magia/forge) delivers a complete voice bot in 12 weeks at $20,000 USD: owner's cloned voice, bay slot integration, deposit collection, operational dashboards. For a single-location shop that wants to start with WhatsApp and add voice later, [MAGIA / Solo](https://catalizadora.ai/magia/solo) at $4,500 USD in 15 days delivers the written agent with CRM and scheduling. Book a 30-minute technical call to review your operation, call volume, and current stack. A real conversation—no pitch deck. ## Preguntas frecuentes ### Can a voice bot schedule an oil change and collect the deposit in the same call? Yes. The bot checks live bay availability, offers three open time slots, confirms vehicle details, sends a payment link via SMS or WhatsApp during the call, and locks the appointment on payment confirmation. Total latency under 800 ms keeps the conversation flowing without robotic pauses. ### How natural does a voice bot sound for a shop in Bogotá? An owner or advisor's cloned voice through ElevenLabs in Latin Spanish is indistinguishable from a human in a 1–3 minute conversation. The trap isn't the voice—it's the latency. If the bot takes 2 seconds to respond each turn, it sounds robotic. With a properly built stack, it responds in under 800 ms. ### How much does it cost to run the bot per month for a mid-size shop? Between $300 and $700 USD per month depending on call volume. For 1,000 monthly calls averaging 2 minutes: ElevenLabs TTS $45, Deepgram ASR $30, LLM $60, SIP telephony $150. Approximate total: $300 USD. Compare that against a 24-hour receptionist costing between $800 and $1,500 USD. ### Can it integrate with the shop's existing calendar? Yes—with Google Calendar, Calendly, or the shop's own system. The bot reads bay and mechanic availability in real time, never books over occupied slots, and respects service type (an oil change doesn't use the alignment bay, for example). If your shop runs on Excel, that works too: it gets digitized as part of the build. ### What if the customer wants to speak to a human? The bot escalates in under 60 seconds. It detects four signals: frustration, out-of-scope question, complaint intent, or an explicit request. It transfers the call to a human advisor with full context pre-loaded. The customer doesn't repeat themselves—the human picks up already informed. --- Source: https://catalizadora.ai/blog/bot-de-voz-para-taller-mecanico-que-agende-servicios-en Author: Pablo Estrada — AI Catalyst, LLC (catalizadora.ai)