SUCCESS CASE

Vehicle
commands by
voice with SLM
and Edge AI at no
cloud cost

We implemented Small Language Models (SLMs) on the mobile device for a major German automaker. 100% processing on the user's hardware via Apple Foundation Model eliminated cloud costs, ensured GDPR compliance, and drastically reduced latency.

Edge AI
GenAI
SLM

Results

Zero operational cost per request with the elimination of cloud infrastructure

Minimum latency with immediate response in vehicle command execution

Full regulatory compliance with Privacy by Design in accordance with GDPR rules

Complete mitigation of leak risks by eliminating sensitive data traffic

Overview

The mobile app of a major German automaker used a cloud-based AI architecture dependent on global servers. The model generated high infrastructure costs, budget allocation disagreements between departments, noticeable latency in vehicle commands, and strict compliance challenges with European data privacy laws (GDPR).

The challenges

Growing cloud infrastructure costs with per-request processing.
Latency in executing vehicle commands due to traffic on remote servers.
Strict GDPR compliance requirements and protection of sensitive customer data.

Journey
and solutions

We migrated the intelligent chat to the mobile environment through Small Language Models (SLMs) integrated with the Apple Foundation Model. Intent processing and tokenization now occur 100% on the user’s hardware (iPhone, Mac, and iPad). The solution interprets voice or text commands and triggers immediate telemetry actions directly to the vehicle, eliminating traffic through external servers.

What did we deliver?

Migration of the AI architecture to the Edge AI environment on the user's mobile device.
Integration of Small Language Models (SLMs) with the Apple Foundation Model.
Local natural language processing for vehicle telemetry actions.
Architecture with a Privacy by Design approach in strict compliance with GDPR.

More success stories

Seguraí: Generative AI Assistant for Policy and Insurance Queries

We developed Seguraí, a generative AI virtual assistant via AWS Bedrock for the insurance sector. The solution uses a private LLM and natural language processing for instant policy queries, ensuring data security and operational efficiency.

Vinci Partners: 100% digital journey in the pension and insurance market

ília led the Product Discovery and end-to-end development for Vinci Partners, creating Vinci Vida e Previdência. The solution eliminated paper-based processes and launched the Mio app and 3 portals, consolidating a 100% digital contracting journey.