Medium: How AI entered a UX designer's daily workflow at a B2B cybersecurity company
Laís Lara Vacco is a UX designer at Cobalt, a B2B cybersecurity company, and her article documents how AI tools entered her working process over several months — what replaced existing habits, what stayed the same, and what became genuinely faster.
Context: a product team at a B2B company
Cobalt builds security testing software, and the design challenges it faces are typical of B2B products: complex domain knowledge, multiple user types, and a codebase that is already in production and needs to be understood before anything new gets designed. That context matters for how AI tools were actually useful, since many AI-design case studies come from consumer apps where the problem space is more familiar.
Research and onboarding
The team started using NotebookLM for internal knowledge management. They uploaded project documentation, asked follow-up questions against it, generated visual maps of concepts, and used it to get up to speed quickly when moving between product areas. The main benefit was time: switching context on a complex B2B product involves a lot of reading, and having a queryable layer over existing documents shortened that process.
Prototyping
The more significant change was in how early the team started working with interactive prototypes. Instead of beginning with static wireframes and moving to clickable prototypes later, they used Subframe, AI Studio, Builder.io, and Figma Make to generate functional prototypes much earlier in the design process. Claude and Claude Code handled text-to-code work within that flow.
This shift did not remove any design steps — it moved them earlier. Getting to something clickable before refining layout or content helped the team identify problems in user flows at a stage where changes cost less time.
What this account offers
The article is a description of one team’s experience, not a universal framework. The tools may not map to every team’s stack, and the B2B context shapes which problems AI actually helped with. It is most useful as a realistic account of AI-assisted design that goes beyond tool lists, showing the specific moments where AI changed the sequence of work rather than just the speed of individual tasks.
Useful for UX designers who want to see how AI integration into a design workflow looks in a complex product context, and for teams considering which parts of research and prototyping to approach differently.