You have no notifications.
I’m sure everyone who attended Kentico Connection Prague 2025 would agree that there were many inspirational ideas and practical tips shared during the event. While many were discussing business and sharing their tip for succeeding with Kentico, there was another important theme that ran through every presentation in the technical track and the many follow-up discussions - agentic software practices.
It began earlier in the day during our Kentico Product Update, where David Slavík’s live demo set the tone. He showed how a technology often labelled “non-deterministic” could be guided - through the provision of structure, context, and orchestration - to perform a range of development tasks reliably and consistently. Watching a set of agents automatically generate components and resolve breaking changes within minutes was, for many in the audience, an eye-opening moment. In fact, it reframed the day’s technical discussions – topics that carried on late into the social event that night. We were no longer talking about whether AI could help, but rather how far it can go in transforming development workflows. The sessions that followed built on one another - from individual productivity to orchestration at the team level, and finally, to autonomous systems capable of self-healing and optimization. Together, they painted a clear picture of where digital experience development is heading and how Kentico is helping partners get there.
The first session, Boosting Productivity in Migration Projects by Milan Lund, explored the practical application of agentic workflows for developers tasked with complex migrations.
Milan’s presentation resonated because it captured something every developer knows: migration projects can be tedious, repetitive, and error-prone. His approach flipped that narrative. Using workspaces, file referencing, and contextual prompts, he showed how developers could operate as solution architects, while AI acted as the “mid-level developer” executing migration steps. I recall him saying that it means “Less typing, more thinking,” for developers going forward, emphasizing that AI and humans working together is a powerful combination.
In his experience, the gains have been tangible: assembling migration prompts in minutes instead of hours, generating structured code aligned with Kentico’s best practices, and iterating rapidly toward production-ready quality. For developers and agencies, this shift offers more predictable delivery timelines, could reduce manual overhead, and give people the space to focus more on innovation rather than rework.
Building on that foundation, Liam Goldfinch took the conversation into broader team collaboration with his talk, “Can AI Really Speed Up Xperience by Kentico Project Delivery?”
He began by pinpointing where projects typically slow down, such as boilerplate setup, code reviews, documentation drift, and constant context switching. His solution wasn’t a single tool, rather he presented more of a mindset shift: combining GitHub Copilot, Claude Code, and Model Context Protocol (MCP) servers to create, what he referred to as, spec-driven development workflows. Liam demonstrated how AI could generate documentation, validate pull requests, and even draft full modules — all while staying grounded in a project’s unique context.
That thread of context carried directly into Łukasz Skowroński’s session, which zoomed out to the infrastructure level. His talk, “Some Fun with Xperience by Kentico,” unpacked how the growing number of MCP servers can become increasing unwieldy to work with as a developer, and he presented a way of organizing and optimizing them using the Docker MCP Toolkit.
Łukasz described how containerization provides both control and safety in increasingly complex AI environments, helping developers to isolate projects, manage multiple agents, and route tasks through a consistent architecture.
His key insight: treat everything like an application.
By treating every AI agent and MCP server as modular yet connected components, his approach could help teams to strike a balance between flexibility and governance, thereby ensuring AI actions remain traceable, reproducible, and project-specific – a tip that’ll be very handy for agencies working on many Kentico upgrade projects, I’m sure!
The final technical session set a course for the future, triggering the imagination of attendees and sparking many conversations afterwards. In “Building an Agentic Future with Xperience by Kentico,” Jeroen Fürst explored how agentic architectures could evolve into self-healing DXPs. He began with a simple question that had enormous impact: “What if your website or DXP could take care of itself?”
Jeroen mapped out a vision where a combination of AIRA, n8n workflows, GitHub coding agents, and MCP servers form a living ecosystem that is capable of detecting anomalies, orchestrating fixes, and deploying updates automatically. I particularly liked his automated issue resolution use case where AI scans event logs, identifies a recurring error, generates a fix through a coding agent, and pushes it to review while remaining within a human-in-the-loop framework to ensure quality and safety. It was equal parts technical feasibility and inspiration.
“Agentic DXPs don’t replace humans,” Jeroen emphasized. “They empower them.”
But what resonated most with me was his vision of marketers, developers, and AI systems collaborating in a continuous feedback loop, illustrating that optimisation is no longer a project phase but an ongoing process that starts from Day 1.
Across all of these sessions, a narrative of faster task completions, context-rich collaboration and continuous optimisation emerged. It’s a direction that aligns well with our ongoing work on Xperience by Kentico, as they are each part of the same goal - helping marketers and agencies move faster, smarter, and with greater confidence in an AI-driven world.
But when I think about it, perhaps the most important takeaway wasn’t really technical at all. It was about mindset. We weren’t talking so much about what tools to use, but rather how to think with AI - maintaining human judgment while embracing automation, letting creativity and structure coexist, and discussing how to build systems that learn, adapt, and evolve (just like the teams who create them).
Debbie Tucek
When we met our North American partners in Burlington for Kentico Connection 2025, one theme stood above the rest: AI is officially moving from being hype to being helpful.
At Kentico Partner Connection 2025 in Sydney, Australia one theme consistently surfaced: how do we bridge the divide between AI skeptics and AI enthusiasts - and move from debate to meaningful adoption? By the end of the event enthusiasm became that bridge.
The Prague sessions and subsequent conversations impressed me greatly. It showed me a technical community ready for the next stage of innovation - one where developers are architects of AI-powered systems, and DXPs become active participants in their own growth and maintenance.
We’re no longer asking, “Can AI help us?” but rather “How far can we go together?”
At Kentico, we’ll keep building the tools, frameworks, and partnerships that make that collaboration not only possible but practical. Because the future of DXPs and this amazing community isn’t just agentic - it’s collaborative, contextual, and human at its core.
VP Product at Kentico. Product team leader, and former Marketing Executive, with more than 10 years’ experience across the product management and marketing spectrum – from startups to large organizations - I’m passionate about designing cloud software products and end-to-end services that customers love to use. My superpowers include my unending passion & enthusiasm for what I do and the problems my products solve, and my ability to project this passion to motivate others, be that in grooming sessions or on the international stage.