mimik Extends SOAFEE Blueprints Beyond the Cloud with In-Vehicle API, AI Inference, and MCP Gateway

Today's vehicles aren't just mechanical machines. They're computers on wheels. Modern cars are software-defined systems capable of autonomous driving, over-the-air updates, real-time diagnostics, and immersive digital experiences. But there's a critical challenge: these advanced features depend heavily on constant cloud connectivity. 

When network connections drop, even momentarily, critical functions can be delayed, control signals disrupted, and safety compromised. For OEMs and Tier 1 suppliers, the question isn't whether to build intelligent vehicles, but how to build them resilient enough to operate independently when connectivity fails. 

The answer is hybrid architecture, where intelligence runs inside the vehicle itself, not just in the cloud. This is being showcased through the SOAFEE Blueprint program. 

This shift toward hybrid, offline-first intelligence also requires a new approach for AI models and software agents to expose vehicle services, which we address through the Model Context Protocol (MCP), as we expand on later in this article and in the video below. 

Watch our complete technical walkthrough:

When critical vehicle functions rely too heavily on the cloud, even a momentary network issue can create serious problems. Safety-critical systems like ADAS can't afford milliseconds of latency. Diagnostic systems that can't process data locally may miss critical alerts. And infotainment features that depend on cloud processing deliver poor experiences during connectivity gaps. 

The reality is that no network—5G, satellite, or otherwise—can guarantee 100% uptime across all driving scenarios. That's why we need hybrid architectures that keep intelligence inside the vehicle. 

At mimik, we've built mim OE (our Agentix-Native Execution Layer). It's a lightweight software platform that turns any Domain Controller, Zonal Gateway, Central Compute or end device into a secure, intelligent node. It enables traditional workloads and AI agents to run locally, discover and collaborate in real time, and operate safely, even when disconnected. 

Key features: 

• Offline-first: Works without connectivity, with cloud as an enhancement
• Zero-trust security: Built-in security for every interaction
• Hardware agnostic: Works across ARM, x86, and other architectures
• Real-time capable: Supports both RTOS and non-real-time systems 

Our approach brings the SOAFEE (Scalable Open Architecture for Embedded Edge) framework to life through mim OE, our operating runtime environment. It runs on generic resources with or without a hypervisor, supporting both real-time and non-real-time operating systems. 

Agents running on mim OE can either: 

1. Implement new features as microservices
2. Act as proxies to existing legacy or modern systems 

This dual capability means you don't face an all-or-nothing decision. You can adopt the agentic-native model incrementally, wrapping existing systems while building new capabilities. 

Four Key Benefits 

1. Interoperability: mim OE microservices integrate seamlessly with other components, enabling OEMs and Tier 1s to adopt our model without re-architecting existing vehicle software.
2. Legacy Re-use: When a function already exists (powertrain, ADAS, HVAC), mim OE can proxy it as an API or agent without duplicating implementation.
3. Accelerated Adoption: Developers familiar with SOAFEE's container model gain an easy path to extend features into our distributed ecosystem.
4. Future-Proofing: Bridge the standardized in-vehicle service world with edge-native agents, delivering both compliance and flexibility for AI-driven innovation. 

One distinctive aspect of our approach is support for typed resources. These are components that can't be pooled because of their location, features, or ownership. For example, a front-right LiDAR isn't interchangeable with the left due to different viewpoints. 

Some typed resources are static (in-car displays, entertainment systems), while others are dynamic (the driver's phone, a home EV charger). 

With mimik's ad-hoc clustering and dynamic discovery, nodes can "snap on" new devices and add capabilities to the vehicle's service mesh. Since vehicles outlive personal devices, it's pragmatic to leverage the owner's newer, higher-quality sensors and screens, especially for cockpit UX. The vehicle operates as a system that interacts with others, expanding and contracting with context. 

We've integrated the Model Context Protocol (MCP) to extend vehicle services into the AI domain. While SOAFEE standardizes how vehicle functions are exposed as services, MCP defines how AI agents can discover, request, and consume those services, and how they safely interact with each other. 

This alignment delivers: 

• AI-Native Access: Treat in-vehicle services as AI tools that agents can call dynamically
• Cross-Domain Interoperability: Enable microservices (ADAS, HVAC, infotainment) to interact with agents outside the vehicle
• Context-Aware Behavior: AI agents negotiate typed resources dynamically (e.g., "use the driver's phone display if available")
• Multi-Agent Future: Evolve toward ecosystems where multiple AI agents coordinate complex behaviors 

In other words: SOAFEE defines the "what" of services; MCP defines the "how" of AI agent interaction. 

Here's how it works in practice: 

In the vehicle, you have high-performance computing running QNX and Red Hat over ARM-based processors. Software stacks like Linux or Android manage service execution. Using protocols like SOME/IP, you bridge legacy container runtimes with the microservices world mim OE implements. 

On mim OE, you have services like heated seats and mKey. This is a key generator for handling secure access. This enables the car owner to generate keys and temporarily distribute access to limited functions (perfect for car sharing, valet services, or family use). 

On the user's phone (running Android), mim OE turns the phone into an aggregator of microservices, exposing all the car's features like heated seats or mKey. 

This demonstrates how mimik's SOAFEE Blueprint enhances vehicle frameworks by extending them with a microservice and agent-native approach. The vehicle becomes a "super gateway," bridging automotive and non-automotive developer worlds, evolving into a dynamic collection of AI agents. 

Standards Alignment: Ensures interoperability across your supply chain without vendor lock-in. 

Faster Adoption: Leverages existing SOAFEE investments while extending them to AI and agent use cases. 

Future Readiness: Provides a clear path for evolving toward intelligent, context-aware systems as AI becomes more central to vehicle functionality. 

Reduced Risk: Wrap legacy systems as services, enabling incremental adoption rather than disruptive transitions. 

If you're interested in exploring how mimik's SOAFEE Blueprint can accelerate your roadmap, we offer: 

• Proof of concept programs to validate technology fit
• Integration pilots for real-world deployment testing
• Joint Blueprint contributions to advance the industry 

Whether you're an OEM, Tier 1 supplier, or ecosystem partner, mimik provides ways to collaborate and build the next generation of intelligent, resilient vehicles together. 

 

Key Takeaways: 

• Hybrid architecture solves cloud dependency, ensuring vehicles operate reliably without constant connectivity
• mimik's Agentix platform transform Domain Controller, Zonal Gateway, Central Compute into intelligent edge node capable of autonomous operation
• SOAFEE + MCP integration provides standards compliance while enabling AI-driven innovation
• Typed resources and dynamic discovery extend vehicle capabilities through connected devices
• Legacy system integration protects investments while enabling gradual migration 

Ready to learn more? Contact us to discuss how mimik can accelerate your software-defined vehicle journey.