Start with the work, not the technology

A useful starting point is a workflow that is difficult to complete today: people searching for information, reviewing material or moving between disconnected steps. The first question is whether AI would help, and where a simpler product or web solution would be enough.

I am Norbert Barna, Product VP. I bring product leadership, experience with human-reviewable AI workflows and hands-on platform development. The scope can focus on discovery, collaboration with your existing team or a defined product build. We should agree those responsibilities before committing to delivery.

Where I can contribute

Discovery and prioritisation. Map the user task, business need, existing constraints and a success measure. Separate an attractive demonstration from a problem worth solving.
AI product and integration collaboration. Define how AI should fit into a workflow, what users need to review, and how errors or uncertain outputs should be handled. Assess the proposed integration with the people responsible for your systems and data.
Custom product and web development. Shape and build the interface and supporting product experience. Clarify content ownership, access needs and operational handover as part of the scope, rather than treating the website as an isolated visual deliverable.

Relevant work, with clear boundaries

Instructure demonstrates AI-product work with explainable suggestions, editable outputs and human review. The case describes collaboration with product, engineering, legal and data-science teams.

Raiffeisen shows product leadership and design-system work in regulated banking, including collaboration with engineering and compliance. It is evidence of regulated product work, not a claim of backend AI integration.

Kineticare is a Hungarian digital-health platform I designed and built, including its content system and member experience. It supports the development offer; it is not presented as an AI deployment. These projects demonstrate different relevant capabilities, not three equivalent integration projects.

A proposed way to work together

Frame the decision. Identify the workflow, users, owner and desired outcome. Review what can be assessed with non-sensitive examples.
Test a bounded approach. Agree a prototype or discovery scope, review criteria and the decisions that must remain with people. Check assumptions before expanding the work.
Define the next commitment. Use what we learn to scope implementation, responsibilities, evaluation and handover. Delivery arrangements depend on your context; this is a suggested approach, not a fixed package or timeline.

What to prepare

The workflow you want to improve and who uses it.
The existing systems involved and who owns them.
The data needed, its sensitivity and current access restrictions.
A practical success measure and any baseline you already have.
The person who can make scope decisions and review the result.

A short, anonymised description is enough for an initial conversation. Please do not send credentials, customer records or confidential datasets before an appropriate way to share them has been agreed.

Discuss your project

Tell me where the workflow breaks down and what your team needs to achieve. We can use that to discuss whether discovery, AI-product collaboration or a product and web build is the right next step.