Skip to content
All case studies

Automotive / Voice-controlled workshop

Hey Bob. An extra pair of hands in the workshop.

AITJE, developed a voice-controlled application for a local garage that serves over a hundred customers a day. Six Android tablets connect mechanics to technical information, work orders and their administrative records – simply by asking.

Hey Bob. An extra pair of hands in the workshop.

Carry on working. Just ask.

Keep your hands on the car.
Bob at the information desk.

Standing under a car, carrying out an inspection and walking over to a computer in between: in a busy workshop, administrative tasks constantly interrupt the work. AITJE made the existing garage processes accessible via voice commands. Bob retrieves information and prepares actions whilst the mechanic carries on working.

6
Android tablets
100+
customers per day
1
local BOSGAME M6

“Hey Bob, it’s Dirk.” The tablet wakes up and Bob responds. The conversation remains open, so that a technician can ask further questions or have the next step prepared.

BOB Workshop assistantCall in progress

Hi, I’m Bob,
What’s going on?

Technical enquiries · Work orders · Parts

Example of how it works. Select a task below to view the conversation.

What technicians ask Bob

From specification to work order.

The same point of contact, linked to different systems. Bob uses the context of the employee, the vehicle and the work order to assign the correct task.

Technical information

“Hey Bob, Dirk here. What is the torque setting for the Golf’s cylinder head on the 2?”

Bob uses the active work order to find the correct vehicle. The app retrieves tightening torques, fluid levels or the location of a part from technical data sources.

Haynes / Autodata + active work order

Update work orders

“Note on the Ford Focus service report that the front tyres still h2-millimetre tread depth.”

Findings are immediately incorporated into a proposal on the relevant work order. Recommendations and additional work, such as replacing the brake discs, can also be recorded during the inspection.

Work order in the garage management system

Time recording

“Time me for the MOT on bridge 1 and start the hours for the timing belt on bridge 3.”

The technician switches between tasks using a spoken command. Bob prepares to stop and start the relevant timers. Waiting for a part? Then the timer can be paused.

Time tracking + employee + work order

Parts & stock

“Add a set of front windscreen wipers for a Tesla Model 3 to the order list for the afternoon delivery.”

The app checks stock levels in the ERP system or adds parts to the order list with the wholesaler. This means the technician does not have to walk to the warehouse or the counter to check stock levels.

ERP / stock + parts wholesaler

Inform the reception desk

“Please let the reception know that Mr De Vries’s car is ready and that they can ring him.”

Bob is preparing a status update and a message for reception. A request to the workshop manager – for example, to check for a leak at the bridge 4 – can also be sent via the tablet.

Work order status + reception interface

Tick off inspections

“Tick the boxes for windscreen washer fluid, lights and tyre pressure to confirm your agreement for the current service.”

Verbal findings are entered into the corresponding fields on a digital inspection form. The technician checks the proposal before the records are updated.

Digital inspection form / MOT checklist
BOSGAME-computer with AITJE-logo as a local computing environment
A dedicated on-site computing environment, connected to the workshop tablets.

On-site speech processing

Six tablets.
One local brain.

As the garage handles sensitive customer and vehicle data, AITJE installed an BOSGAME M6 at the company. The six tablets are positioned throughout the workshop and connect to this computer via the local network.

On the BOSGAME, Whisper runs locally to convert spoken commands into text. The processing layer translates the command into an action and sends it back to the tablet app. There, the API connections link to the garage software and technical databases.

Whisper listens. The software carries out the task.
Speech-to-text, recognising staff and invoking tools are separate components of the solution.

This is how an assignment works

  1. Hey BobOne of the six tablets
  2. BOSGAME M6Local transcription & action proposal
  3. The tablet appTargeted API and tool calls
  4. The correct sourceHaynes, Autodata, GMS or ERP

The information found or the proposed action reappears on the tablet whilst the conversation continues. Speech is processed locally; information from external databases is retrieved via their links.

Own hardware. No external AI meter.

How much does a consultation with Bob cost?

One-off hardware€2,000.00

Whisper runs on the BOSGAME M6. Speech and actions are processed locally, without any external charges per audio minute or AI token. A task carried out via the company’s own network therefore incurs no ‘OpenAI’ charge. The links to garage software and technical databases remain part of the solution.

The hardware cost a one-off fee €2,000.00. The external AI costs are €0.
For the ‘BOSGAME’ and all six tablets combined, the power consumption in this calculation example amounts to €6.02 per month. These are hardware and usage costs; development, maintenance and software licences are not included.

Estimated monthly costs for this usage · Whisper transcription + text model for actions
SolutionInput / 1MOutput / 1MVoice / monthText model / monthTotal AI / month
Whisper + local processing$0$0$0$0€0 + power
Whisper API + GPT-6LunaUS$0.10US$0.50US$35.64US$9.50€40.63
Whisper API + GPT-6.1SolUS$2.00US$10.00US$35.64US$190.08€203.15
Calculation example · adjust usage

Six mechanics, 5 –8 cars per mechanic per day and 15 –40 jobs per car. That is approximately 450 –1,920 assignments per day. The cars here are calculated per mechanic; this is an estimate based on usage, not a reading from the logs.

Assignments per month
23,760
Audio minutes per month
5,940
Local tool calls per month
47,520
Adjust conversation duration, tokens & data usage

All figures below are estimates. Audio time is the total duration of the recording submitted per assignment, including follow-up questions. Tokens comprise all model runs, context and tool results combined.

Lower recurring computing costs€199.69/ month

In this calculation example, after deducting the power consumption of the BOSGAME.

Cloud: AI + tablets
€205.71
Local: power supply for the entire setup
€6.02
Annual difference
€2,396.30
Hardware payback period after
10 months

The full ‘€2,000’ hardware is offset against the monthly difference in computing costs. Development, maintenance and existing software licences are not included in this.

A compact local computer

The linked BOSGAME M6 features a Ryzen AI 9 HX370, Radeon 890M, 32 GB DDR5 and 1 TB SSD. For the CPU, BOSGAME states a TDP of 45–65 W. This is not a measured power consumption figure for the entire computer. The calculation uses 50 We are active and 5 Working hours, with eight active hours per working day and a month of thirty days.

The six tablets come together at 8.5 kWh per month; the BOSGAME at 11.5 kWh. Tablet power consumption is counted on both sides.

BOSGAME-specifications ↗

Plenty of promotions, targeted links

1,080 tasks per day, with this example 2 Local tool calls per assignment. These tool calls do not incur a ‘OpenAI’ token charge. Subscriptions or charges with Haynes, Autodata and other linked suppliers are excluded from this.

The cloud comparison uses standard rates, excluding caching or volume discounts. Spoken responses via a paid text-to-speech service, taxes and any additional services are not included. The exact local Whisper format and action model are not specified here; this is a cost comparison, not a test of equivalent model quality.

Rates checked on 4 October 2026: Whisper · $0,006 per minute en OpenAI model rates.

The technician is in charge

Bob is making preparations.
You say ‘OK’.

Before data is written to the administrative system, a final check is carried out. Bob submits the proposed change. Only after the employee has given their approval is it recorded.

Local voice recognition usually helps to identify the right employee for a task straight away. The system logs who carries out which action, so that changes remain traceable.

Example · Ford Focus work order

Front tyres: 2 mm tread depth.
Note: to be replaced shortly.

Staff member: Dirk
“Right, Bob, just put it on the work order.”

Receive approval → update work order → log action

What ‘AITJE’ delivered

An assistant that works with you.

An application running on six tablets, local speech processing on the BOSGAME and connections to the systems the garage already uses. Retrieving technical information, recording findings and informing reception become part of the conversation whilst working.

Fancy a workflow for your work too?

MOVING FORWARD TOGETHER WITH AI

Your next step begins with a question.

Got an idea, facing a challenge, or simply curious? Tell us what you want to achieve. AITJE will work with you to find a solution.