Many people still think DevOps is only for web applications and cloud services. But the reality is clear: companies that apply DevOps principles to embedded systems are outpacing their competition. In this talk, which I gave at a DevOps Meetup in Munich, I explore why embedded teams need DevOps and how to build a Digital Factory that enables continuous value delivery, even for hardware products.
The Broken Value Stream#
In most organizations I consult, the picture looks the same. Business creates plans, writes Jira tickets, and throws them over the “wall of confusion” to Development. Development implements something and throws it to Testing. Testing finds mismatches but signs off anyway. Operations receives the result and says “this will never work in production.” And when the customer finally sees it, they say: “This is not what we wanted.”
The root cause is the silo organization. Business, Development, Testing, and Operations work on different goals. They are not aligned, and the value stream is completely broken by these walls of confusion. The result: inflexible processes, slow delivery, and frustrated teams.
“DevOps is a mindset, a culture, and a set of technical practices which allows us to organize across the value stream and bring all the people together to work on a product.”
From Projects to Products#
Understanding the difference between project thinking and product thinking is fundamental to DevOps. A project has a start point, an end point, and a fixed set of deliverables. It is about maximizing the number of features delivered. Many companies claim they are agile, but they are still working in this project mode.
Product thinking is fundamentally different. In product mode, the customer is at the center. You work on that one feature that solves the customer’s problem, not on ten features because someone wrote them into a project plan. DevOps enables this shift because it brings all the people, processes, and technology together to continuously deliver value along a unified value stream.
Tesla: DevOps in the Real World#
If you think DevOps does not apply to hardware, consider Tesla. In October 2021, Elon Musk tweeted about rolling out version 10.2 of the Full Self-Driving Beta to 1,000 owners with a perfect safety score of 100/100. What does this tell us?
Tesla’s software is modularized. They can update individual modules over the air. They do canary releases, targeting specific user groups. They calculate safety scores through observability. Eight days later, version 10.3 rolled out to a wider group. When a problem appeared, Tesla did a rollback and communicated that this was expected. Less than 24 hours later, they deployed 10.3.1 as a fix-forward.
This is canary releasing, rollbacks, fix-forward, and observability, all in a regulated environment with cars driving on public roads. The companies adopting Lean, Agile, and DevOps in hardware will soon dominate the market, just like Netflix, Google, and Facebook did in software twenty years ago.
The 24 Key Capabilities from Accelerate#
The book “Accelerate: The Science Behind DevOps” identified 24 key capabilities that distinguish high-performing organizations. These are grouped into five areas:
Continuous Delivery: Version control for everything (not just source code), automated deployments, continuous integration, trunk-based development, test automation, test data management, shift-left on security, and continuous delivery to staging.
Architecture: Loosely coupled architectures and empowered teams that can make their own decisions.
Product and Process: Customer feedback loops, value stream organization, small batches, and room for experimentation.
Lean Management: Eliminating harmful change approval boards, monitoring, proactive notifications, WIP limits, and visualizing work.
Culture: Generative organizational culture where finding bugs is celebrated as learning, support for continuous learning, cross-team collaboration, job satisfaction, and transformational leadership.
The research shows that high performers achieve 973 times more deployments to production compared to low performers. Companies doing DevOps have faster time to market, more value for money, higher quality, higher customer satisfaction, and top qualified employees.
Continuous Testing and Quality#
Joe Justice, an Agile coach and former Tesla employee, stated that 50% of the money a Musk company invests into a new product goes into automated testing. In most companies I see, only a fraction goes into testing. This accelerates development at first, but eventually progress stalls because quality degrades.
The shift from the traditional V-model to modern testing is dramatic. In the old world, months could pass between writing a feature and testing it. With Behavior Driven Development (BDD), we define acceptance criteria in “given, when, then” form and bake quality in from the start. With Test-Driven Development (TDD), we write tests before code. The traditional test pyramid gets flipped: many unit tests (fast and cheap), some integration tests, and only a few end-to-end tests. The focus shifts from “find every bug” to “prevent bugs” through a risk-based approach.
“In the traditional model, the focus is on finding every bug, which makes it very slow. In the Agile test pyramid, we want to prevent bugs. It is a risk-based approach.”
Platform Engineering: Scaling DevOps#
When teams practice DevOps, each one needs infrastructure, CI/CD pipelines, monitoring, security tools, and access management. At scale, this creates massive redundancy, inconsistency, and cognitive overload. Each team reinvents the wheel.
Platform Engineering solves this. A dedicated platform team builds and maintains a standardized platform containing all the tools teams need: application runtime, developer experience, automated DevSecOps pipelines, access and identity management, and observability. Product teams still own and operate their products, following the “you build it, you run it” principle. The platform team simply provides the foundation.
The important distinction: customers do not pay for your platform. They pay for your products. But the platform generates value for your teams, enabling them to build better products faster.
Building a Digital Factory#
The Digital Factory is the holistic framework that ties everything together. At the portfolio level, the board of directors aligns big ideas with the company vision. At the product level, product management breaks down epics into features. At the team level, empowered teams work in an agile way on their modules, supported by a standardized platform.
This creates a continuous loop: fast delivery of features to the customer, fast feedback from the customer, and continuous learning. Platform Engineering builds the foundation. DevOps teams practice on top of it. And the Digital Factory orchestrates the whole system.
“Platform Engineering builds your platform, which is the foundation of your Digital Factory, and enables your teams to do DevOps.”
Key Takeaways#
Products over projects. Stop maximizing features and start solving customer problems. DevOps enables the shift from project thinking to product thinking.
Quality is not optional. Invest in automated testing from the start. Use BDD, TDD, and a risk-based test strategy. The cost of skipping tests always catches up with you.
Platform Engineering enables DevOps at scale. A standardized platform eliminates redundancy, reduces cognitive load, and lets product teams focus on delivering value.
Build a Digital Factory. Connect strategy with execution through Lean Portfolio Management, Platform Engineering, and empowered DevOps teams. This is how you industrialize embedded software development.
Learn from Tesla. Even in regulated, hardware-heavy environments, DevOps practices like canary releases, rollbacks, and observability are not just possible, they are a competitive advantage.