AIGU
1 | ABOUT THE PROJECT
AIGU's goal was to enable drivers to clear their email inboxes, structure their thoughts, or log CRM entries hands-free immediately after a meeting, before key details fade. As the technology partner, we at MeguMethod managed the entire product lifecycle: from the initial Product Discovery and robust architecture design to the final integration into enterprise systems.
20+ interviews
Target audience research confirmed the frustration of forgetting ideas behind the wheel.
900 CZK
Validated monthly price acceptable to companies as a business expense.
Resource Allocation
80% development (native iOS, server, real-time AI) vs. 20% pitch and business plan.
2 | CHALLENGE
Building a voice interface for a moving car meant addressing extreme demands on latency, connection stability, and audio clarity. We successfully built the entire system using cutting-edge technologies.
Native iOS Application
To ensure 100% smooth performance while driving and deep integration with Apple CarPlay, we opted for purely native development.
Node.js backend & Audio Tunnel
We developed a robust Node.js backend that acted as a stable bridge for real-time audio streaming.
OpenAI Realtime API
We tunneled the audio signal from the car through our server directly to the OpenAI Realtime API. This allowed the assistant to process spoken language instantly and respond without frustrating delays.
Alternative Prototypes (Phone-Based AI)
As part of the validation process, we also developed and tested an alternative option – a voice assistant on a standard phone line. The system worked in a way that the assistant could either call you directly, or you could dial the number to speak with it. From a technological perspective, it was an excellent and fully functional solution that confirmed our ability to integrate AI into any communication channel.
If I don't handle something right away, I forget it. Those little things—jotting down a task, adding a meeting to the calendar – if I don't do them immediately, they just don't get done. If the assistant saves me ten percent of my time, it equals the cost of a daily coffee.
3 | FINDINGS
This project provided us with unique technological and product know-how, which our clients seeking similar solutions directly benefit from today.
01
Proven Architecture Ready to Deploy
The entire technical solution—from the Node.js audio tunnel to the OpenAI Realtime API integration—is complete, tested, and fully functional. If you are looking for a similar voice or AI solution for your business, you don't need to start from scratch. We have a proven foundation readily available.
02
Voice UX Has Its Specifics
We learned how to clean the audio of filler words ('um,' 'actually') and filter out highway noise before the sound even reaches the LLM models.
03
Confirmation Bias and Trimming the MVP
We know how easy it is to fall in love with your own product. Today, we know that risk analysis and trimming the MVP down to the absolute minimum (Minimum Viable Prototype) must precede writing a single line of code.
AIGU was our internal startup project. Despite successfully mastering it technologically, we decided to discontinue it for the commercial market in the fall of 2025. We invested over 500,000 CZK and months of work into it, only to hit the business limits of the venture capital market at the time and a market niche that was too broad. Why are we mentioning this? Because we paid out of pocket to explore all the dead ends in real-time voice AI. We won't be repeating these expensive mistakes on your client projects.
Want to implement AI or voice without the risk?
We believe the future lies in voice interfaces—be among the first to adopt voice as one of your interfaces.