THE AI STARTUP - ENGINEERING COMPANY - ASCS Brochure #2025 - Magazine - Page 45
THE AI STARTUP - ENGINEERING COMPANY
THE AI STARTUP
ENGINEERING OFFICE
Mingabyte was founded in 2021 with a clear vision: to develop complex,
safety-critical products more intelligently—while radically simplifying the
associated bureaucracy. Today, the company combines the agility of a startup
with the expertise of a highly specialized engineering firm, leveraging AI to
automate time-consuming engineering processes. We spoke with founder
Raimund Waning.
What mission defines Mingabyte, and what inspired
you to start the company?
It wasn’t a single event that led to the founding. It was more
of a process, shaped by reflection and planning. I’ve been
fascinated by seemingly intelligent, highly automated
machines since childhood and always knew I wanted to
work in this field. But another big driver was growing
frustration after years in engineering and as a safety expert
- a feeling many colleagues share. We’re engineers; we want
to be creative and build great products. Yet we’re constantly
slowed down by processes, politics, documentation
overload, or clunky tools. How often does “agility” turn into
chaos, and “compliance” into a bureaucratic checkbox
exercise?
With Mingabyte, we set out to shift the focus: to look
beyond the day-to-day grind and tackle these widespread
pain points head-on. For a long time, Europe’s economic
strength has masked many inefficiencies in development.
But now, competitors from China and India are often faster,
more cost-effective, and just as good in terms of quality.
That’s why we’re pursuing a twofold mission:
On the one hand, we offer product development and
management consulting in strategy, processes, safety, and
reliability for complex systems. On the other hand, we build
smart, scalable solutions that genuinely lighten the load of
daily development work and measurably boost
productivity.
That sounds ambitious.
It is - especially for a bootstrapped family business. For
instance, finding exceptional talent in systems engineering,
process automation, software development, and industry
standards isn’t easy. That’s why training and education are a
core part of what we do.
But these areas don’t operate in silos - quite the opposite.
Successfully simplifying and automating development
processes requires a deeply interdisciplinary approach.
Since we develop sophisticated systems and ECUs all the
way to production readiness ourselves, we know exactly
where and how process simplification, smarter tools,
optimized language models, or AI agents can add real value
- without introducing new headaches.
We take an unconventional approach to development
processes - in the best sense of the word. One example:
instead of manually analyzing thousands of requirements,
we train language models that “understand” the project
context and deliver usable draft outputs quickly. Even
architecture design drafts and test cases can be generated
with minimal manual effort.
We’d rather spend our time on creative work than on
documentation and keep the experts in the loop where
they are still required and provide the most value. And
thanks to our group partners, Wertefest in Munich and
requisimus in Esslingen, we’re able to scale for large
projects and respond flexibly to staffing or expertise
needs.
How do you use simulation technology at Mingabyte?
In developing AD/ADAS systems, we use simulation to run
scenario- and data-driven refinement and verification.
Simulation lets us create a digital twin that allows us to
observe, optimize, and validate AI-driven behavior-specifically control algorithms, not language models - in realistic
scenarios and edge cases. We generate these scenarios
generically and partially automate the creation of requirements, scenarios, and test cases to cover as many edge
cases as possible.
Several of our team members have conducted foundational research in this area, which we continue to build on in
collaboration with leading universities.
I’m convinced we’ll see more AI in safety-critical domains.
But because of its black-box nature, we’ll need new trustworthy ways to validate it. Simulation will play a key role in
making that possible.
www.mingabyte.de
45
