AI is speeding up software development, but Craft Conference 2026 made one thing clear: security, code review, and human judgment matter more, not less, once AI enters the workflow.
This June, four levi niners attended Craft Conference 2026 in Budapest, one of Europe’s leading gatherings for software engineers, architects, and technology leaders. Across dozens of talks, workshops, and conversations, one topic consistently surfaced: how software teams can embrace AI without compromising quality, security, and sustainable engineering practices.
For Anja Basara, Software Engineer, the conference was valuable not only because of the technical content, but also because of the people and discussions surrounding it: “I really liked the atmosphere of the event, lots of networking, diversity of topics and a lot of passionate developers, all with the same challenge – how to survive and thrive in the age of AI.”
Key takeaways at a glance
- Security has to be built into AI-assisted development from day one, not bolted on afterward.
- Third-party packages and NPM dependencies need real scrutiny before they enter a codebase touched by AI tools.
- AI accelerates output, but humans stay accountable for reviewing and validating what it produces.
- Faster code generation can quietly inflate technical debt through longer pull requests and “code review fatigue.”
Why Security Is a First-Class Concern in AI-Driven Development
Among all the sessions she attended, one stood out the most: “Cyber Attacker Mindset for Developers: Secure Software Delivery in the age of AI” by Yossi Sassi.
As Anja Basara explains:
Session held by Yossi Sassi stood out to me the most because he provided concrete examples on how to develop securely in the AI era and how with great power comes great responsibility. We need to consider the attacker’s mindset in software delivery, therefore we need to embrace the best practices for agents, LLMs and vibe coding, to minimize and control data exposure. Also, we need to carefully evaluate third-party packages before using them because they can be very vulnerable to attacks.”
The topic is becoming increasingly relevant as organizations integrate AI-assisted development into everyday workflows. While AI helps teams move faster, the responsibility for secure software delivery remains firmly in human hands. The conference highlighted the growing importance of understanding software supply chain risks, validating dependencies, and incorporating security thinking from the earliest stages of development.
Three Insights Worth Bringing Back
Asked about her biggest takeaways from Craft Conference, Anja identified 3 insights that resonated throughout the event:
- Human collaboration and brainstorming remain essential despite AI.
- Third-party dependencies and NPM packages require careful security checks.
- AI should accelerate development, but developers remain responsible for reviewing and validating the output.
Together, these point to one message: AI is transforming software development, but engineering excellence still depends on human expertise, collaboration, and accountability.
One lesson that particularly changed Anja’s perspective was the realization that effective AI adoption requires a stronger focus on responsibility and security, not less.
“The Cyber Attacker Mindset session made me realize that using AI effectively also means using it responsibly. Instead of assuming AI-generated code is safe, we should validate its output, review dependencies carefully, and think about security from the start of the development process.”
This shift is especially important as more organizations move from experimenting with AI tools to incorporating them into production software delivery processes.
What Matters Most for Customers
For Levi9’s customers, the most important takeaway isn’t a new tool or technology, it’s the mindset needed to use AI responsibly while maintaining trust, quality, and security.
As Anja puts it:
“The most relevant insight for our projects and customers is the need to use AI responsibly and securely. As AI becomes a bigger part of software development, we need to ensure that generated code, dependencies, and third-party packages are properly reviewed instead of being trusted blindly. Adopting a security-first mindset, thinking like an attacker, and validating the tools and packages we use will help us build more secure and reliable products for our customers while reducing unnecessary risks.”
For businesses investing in digital products, this means that successful AI adoption is not simply about increasing development speed. It is about balancing productivity gains with governance, quality assurance, security practices, and engineering discipline.
Looking Ahead: Productivity vs. Technical Debt
Craft Conference also surfaced some of the less-discussed consequences of widespread AI adoption. Referencing ideas from Gergely Orosz’s session, “Slow down to speed up,” Anja pointed to a challenge many engineering teams are already running into:
“AI already makes more tech debt, creates more, longer pull requests, making developers overwhelmed with the amount of code which leaves developers with ‘code review fatigue’ which can lead to burnout. We need to make sure it does not come to that and keep human interaction at work and during development. AI will continue to increase developer productivity, but teams will need better review practices, stronger collaboration, and a greater focus on code quality and security to avoid accumulating technical debt.”
As AI becomes a standard part of software delivery, organizations will need to adapt not only their technology stack, but also their engineering practices. Better review processes, thoughtful collaboration, and a continued focus on quality will be essential for sustaining long-term productivity.
Final Thoughts
Craft Conf 2026 reinforced a message that resonates strongly with how we work at Levi9: AI is becoming a powerful accelerator, but successful software engineering still depends on people.
The future belongs not to teams that automate everything, but to teams that combine AI capabilities with strong engineering fundamentals, secure delivery practices, and meaningful human collaboration. And according to the conversations and insights shared at Craft Conf, that future is already taking shape.
In this article:
