Automotive After-Sales Virtual Assistant: The Authos Case Study
3,364 workshop calls handled in less than two months by “Marta,” Authos’ DeepAgent-powered inbound virtual assistant. How did they achieve a 34.2% first-contact resolution rate for complete cases?
Customers calling a workshop often have an urgent problem and limited information. An assistant that collects all necessary details on the first attempt avoids bounces and callbacks. In this case study, we demonstrate how Authos S.p.A. Società Benefit introduced “Marta,” an automotive after-sales virtual assistant built on DeepAgent, to manage inbound calls. In less than two months: 3,364 calls managed, 160 hours of conversation, 1,671 structured requests for the workshop, and a 34.2% rate of complete and confirmed cases. ## TL;DR - 3,364 calls managed in 53 days; daily volume increased ×3 from August to September. - 1,671 structured workshop requests; average conversation 2m52s. - 34.2% complete and confirmed cases; 38.3% for classifiable calls. - 4-step flow: qualification, data collection, management, non-skippable summary. - Workshop protected: no promises on slots/prices/coverage without verification. !Automotive After-Sales Virtual Assistant: The Authos Case Study — figure 1 ## The Client: Authos S.p.A. Società Benefit Authos S.p.A. Società Benefit has been operating in mobility for over 60 years with locations in Turin, Moncalieri, and Ciriè, supporting customers throughout the car's lifecycle: sales, rental, workshop, and after-sales assistance. The objective of the project with DeepAgent: to transform fragmented calls into actionable cases on the workshop's first callback. ## The Context and Starting Point Phone inquiries range from service appointments and diagnostics to updates on ongoing work, quotes, and manufacturer technical campaigns. Missing data (license plate, mileage, warranty, location) generate further calls. Many dealerships are considering an AI receptionist for dealerships to prepare work for the workshop without replacing it. ## The Challenge: From Incomplete Request to Actionable Case Objective: Deliver ready-to-use cases to the workshop, reducing callbacks for data integration. - Quickly determine if it's maintenance, work status, a quote, a recall, or an out-of-scope request. - Distinguish between ownership, rental, and leasing, with different processes and authorizations. - Collect essential data without turning the call into an interrogation. - Avoid making promises: no confirmation of appointments, prices, or coverage over the phone. ## How the Automotive After-Sales Virtual Assistant Works 4-step conversational flow, designed and iterated with the Authos team. ### 1) Welcome and Routing Marta introduces herself as Authos' after-sales assistant, listens to the reason for the call, and routes it to the correct path. ### 2) Case Qualification Identifies ownership, rental, or leasing; collects data one by one: license plate (or model/year), registration, mileage, owner, declared warranty, preferred location, and urgency. ### 3) Request Management Records technical issue, desired day and time slot without presenting them as confirmed. For rental/leasing, notes the company and clarifies that authorizations and coverage will be verified by the team. Always asks about the need for a courtesy car. ### 4) Summary and Handover to Workshop Reads back the case; the customer confirms; Marta explains that the workshop will call back to confirm details. A service appointment booking over the phone thus becomes a complete case ready for verification. !Automotive After-Sales Virtual Assistant: The Authos Case Study — figure 2 ## The Results: 3,364 Calls in Less Than Two Months Increased scale without sacrificing quality: in September, daily calls tripled compared to August (≈30 → over 90), while the proportion of complete cases remained stable (35.8% → 33.7%). | Indicator | Result | |---|---| | Calls Handled | **3,364** | | Conversation Time | **160 h 14 m** (average **2 m 52 s**) | | Workshop Requests | **1,671** (49.7% of calls) | | Information Requests | **892** | | Complete and Confirmed Cases | **1,149** · **34.2%** | | Complete Cases among Classifiable | **38.3%** | | Calls Transcribed | **99.6%** | Further operational details: - Unclassifiable calls: **368** (average duration **15 s**: dropped calls/silence/abandonments). - Calls per day: **29.9 → 93.4** (×**3.1**). - August: **748** calls; September: **2,616** calls; quality virtually stable. ## The Data to Read Carefully: The 34.2% The 34.2% doesn't mean that one in three calls “succeeds” and the others don't: it indicates cases that are complete in all aspects — vehicle, urgency, location, availability — with a summary confirmed by the customer. A single missing element or a call ending before the summary prevents it from being counted, even if it's still useful to the workshop. For classifiable calls, the figure rises to 38.3%. ## The Conversation that Explains the Method The best call of the period: a service request for a privately owned car, **4 m 08 s**. Marta identifies the need, collects and confirms the license plate, registration, mileage, owner, and warranty; records the preferred location, day, and time slot (without confirming the slot), asks about a courtesy car, and concludes with a summary and handover to the workshop. The automatic evaluation assigns **10/10**. ## What “Marta” Covers in After-Sales Managed cases: service appointments, diagnostics, software updates and ECU checks, work status updates, courtesy car requests, and vehicle pick-ups. The assistant also handles vehicle recalls: it collects data from the letter/technical campaign and forwards it to the workshop for verification and scheduling. !Automotive After-Sales Virtual Assistant: The Authos Case Study — figure 3 ## Lessons from this Case Study - Non-skippable summary: this is the main lever to increase complete cases. - Prudent promises: no confirmation of slots, prices, or coverage without workshop verification. - Separate the “noise”: measure dropped calls and abandonments separately to read the real success rate. - Quick path for urgent inquiries: those calling about an open ticket should be directed to an operator. - Callback capability: if volume triples, the team must be able to call back quickly. ## The Workshop Receives Cases, Not Just Phone Calls The value of an automotive after-sales virtual assistant lies in delivering ready-to-work cases. For Authos, this means **3,364** calls managed in less than two months, **1,671** structured requests, and stable quality even with tripled volume. Also read the case study on the AI WhatsApp agent for car dealerships (sales department) for an end-to-end view. ## Conclusion With DeepAgent, Authos has transformed after-sales inbound inquiries into actionable cases, protecting workshop time and quality. The model sustains increased volume without compromising the completeness of collected data. Want to see how it would work in your context? Request a demo and test the agent in real-time. ## Frequently Asked Questions ### What differentiates a voice assistant for after-sales from a traditional IVR? An AI-powered voice assistant understands intents and language variations, guides data collection, and produces a complete case, not just a simple transfer. It integrates protocols for ownership/rental/leasing, manages exceptions, and concludes with a confirmed summary. Compared to an IVR, it reduces callbacks and lost workshop time, improving quality and throughput. ### Can we start only with the “service appointment” use case and then extend? Yes. Many projects begin with service appointment booking over the phone because it has high volumes and a clear data schema (license plate, mileage, warranty, location, availability). Once quality is validated, it can be extended to diagnostics, work status updates, quotes, and technical campaigns, while maintaining prudent rules on promises and confirmations. ### How does it integrate with internal workshop systems? The assistant can operate stand-alone with structured tickets via email/CRM or integrate via API with DMS and workshop planners. Integration allows for pre-populating customer records, validating vehicle data, and orchestrating callbacks. You can start “light” and connect in subsequent phases, maintaining audit trails and quality metrics. ### Is it suitable for multi-location groups and rental/leasing vehicles? Yes. The flow handles multiple locations, preferred times, and distinguishes between ownership, rental, and leasing, clarifying that authorizations and coverage are verified by the team. For multi-location groups, cases are routed to the correct location, improving workload distribution and reducing waiting times. ### Which KPIs should I monitor to assess ROI? Focus on: rate of complete and confirmed cases, percentage among classifiable calls, average conversation time, volume by type, callback times and rates, percentage of valid transcriptions. Compare before/after on avoided callbacks and staff time transferred to the workshop. These KPIs guide tuning and scalability. Talk to sales → · Create your agent →