Akkodis deploys multilingual AI at Mercedes-AMG PETRONAS Formula One team races
Akkodis has deployed a multilingual AI solution that gives international guests in the Mercedes-AMG PETRONAS Formula One Team Viewing Gallery real-time translated team radio, race commentary and event information.
- under 50 milliseconds
- more than 4,000 TOPS
- sub-600W
- more than 50
What Happened
Akkodis, a global digital engineering consulting company and part of the Adecco Group, announced the deployment of a multilingual AI solution for the Mercedes-AMG PETRONAS Formula One Team Viewing Gallery. The system lets international guests access real-time translated team radio and race commentary, event information and Formula 1 communications in their preferred language. It is built using Akkodis' AI-Core system and processes information directly at the circuit to deliver responsive, low-latency transcription and translation. The solution debuted at the Formula 1 Italian Grand Prix in Monza.
Under 50milliseconds
For real-time speech-to-text transcription of team communications processed locally at the circuit.
- Dedicated AI-Core speech-to-text models transcribe communications in real time at the circuit.
- Domain-specific models recognize F1 terminology, driver names, team personnel, locations and event-specific vocabulary during transcription.
- An AI-Core dedicated Expert Language Model translates the information simultaneously into multiple selected languages.
- More than 50 languages are currently supported.
- A dedicated Terminology Manager allows race-specific terminology to be updated, validated and incorporated into future model enhancements as it changes through the season.
Beyond software and models, Akkodis engineers designed and built the custom physical hardware that powers the solution at the circuit. The compact compute platform delivers more than 4,000 TOPS of advanced AI processing performance within a sub-600W power target. The enclosure, mounting architecture, ventilation and cooling systems were engineered to operate reliably in the demanding F1 garage environment, where high temperatures and limited airflow can challenge performance. Akkodis teams developed prototypes, validated continuous-operation performance and integrated the solution into the live race environment; throughout race week the platform operated continuously and reliably without interruption.
“F1 is an environment where speed, precision and performance matter in every moment. The greatest value from AI comes when intelligence is available where information is generated and decisions are made. This deployment demonstrates how Akkodis Intelligence brings together advanced technology, engineering expertise and human insight to deliver the right information, at the right time and in the right context. By combining AI-Core with purpose-built infrastructure, we're helping create a more engaging and accessible experience for international guests in one of the world's most demanding environments.”
“We always strive to deliver the best experience for our race guests – and bringing leading technology to the fore is a key component. The garage is not only a demanding environment; it also offers an enthralling insight into how our team operates at the leading edge of sport and technology. By working with Akkodis, combining leading language models with bespoke hardware, we've brought our guests closer to the action than ever before. Speed, accuracy and consistency are crucial in this deployment, and we've been delighted with our guest response.”
While first deployed in Formula 1, Akkodis says the architecture demonstrates a broader shift toward intelligent edge environments, showing how advanced AI, local processing, Physical AI infrastructure and engineering expertise can combine to deliver real-time intelligence where speed, reliability and contextual understanding are critical. The project brought together expertise across AI, software engineering, hardware development, integration, systems engineering, mechanical design and deployment operations. AI-Core, Akkodis' modular AI ecosystem, is built for cloud, hybrid and edge environments and has already been validated on leading platforms such as Microsoft Azure.
Why this matters
The deployment shows AI moving to the edge: the system runs directly at the circuit rather than in the cloud, cutting translation latency to under 50 milliseconds for guests who would otherwise follow race communications in English. Akkodis says the approach points to a broader shift toward intelligent edge environments, though it has so far been proven only in Formula 1.
Terms in This Story
- TOPS
- Trillions of operations per second, a measure of how much AI compute a processor can perform.
- Low latency
- A very short delay between an input and the system's response, so results appear nearly instantly.
- Edge
- Computing that runs locally where data is created, rather than sending it to a distant cloud data centre.
- Speech-to-text
- Technology that automatically converts spoken audio into written text.
Summarised from the linked release; details can be imperfect — always verify against the original source.