:quality(80))
Michal Nowakowski advocates for redefining software development in the light of AI, introducing not only disposable software as a way to minimize risk, but also spotlighting the emergence of data as a central player.
Michal proposes a paradigm shift from attaching high value to every line of code to treating software as disposable, enabling quick, low-risk validation of new ideas. He highlights the role of AI in significantly reducing the cost of being wrong, thereby encouraging more experimentation and courage to question assumptions.
Hear Michał highlight:
The idea of disposable software, building lightweight solutions to quickly test and validate ideas while keeping the cost of failure low.
How AI empowers teams to experiment more, run broader test sets, and move from concept to implementation faster.
Why a strong AI strategy should emerge from team-led ideation, rather than a narrow focus on shipping a single AI-powered feature.
The untapped value of user-generated data and its role in strengthening AI-supported product development.
A workshop mindset that starts with imagining unlimited resources, then narrows down to what's realistically achievable.
The importance of treating every initiative as a hypothesis, applying disciplined testing methods inspired by data science.
How validation is evolving—from slow, expensive, and bias-prone processes to faster, more accessible experimentation that actively challenges assumptions.
Quote
Monterail Team Analysis
Here's how software teams can reformulate their approach based on the insights provided:
- Embrace the concept of disposable software: Allow room for more experimentation on ideas and understand that not all lines of code are indispensable. Encourage failure and value the learning opportunities it presents.
- Re-strategize ideation process: Instead of focusing on a quick AI feature, build a comprehensive AI strategy with the team. Explore how software development can be tailored to effectively utilize user interaction data.
- Leverage user-generated data: Unleash the full potential of data generated by user activities. Treat this data as an opportunity to learn about the user interaction patterns and to make data-informed decisions.
- Augment your brainstorming session: Initiate workshops with the vision of unlimited resources, then strategize to maximize the useful parameters based on feasible limits. Dream big, then work backwards.
- Transition from traditional proofing: Borrow strategies from data science. Treat your ideas as hypotheses and create test conditions to validate them. Leverage AI to make validation infeasible both in terms of time and cost.
- Dissociate code validity from AI component: Remember, just because an AI component works with your code, it doesn't necessarily mean your software is optimal or your hypothesis is right. Keep open to questioning and disproving your assumptions.
- Learn from failed validation: See a failed validation of an assumption not as a setback but as a learning opportunity. Remember, the alternative cost is much lower now with AI.