Whiteboards with CLI

There is a very simple way to measure a designer's seniority: analyse his or her dependence on tools. When starting out, it's natural to become obsessed with the latest trendy software, the latest methodology or the latest set of prompts which guarantees effective results in seconds. If I had to give just one piece of advice to anyone starting out, apart from a good library, I would say never leave a whiteboard behind. It is in those diagrams, arrows and shapes that look like hieroglyphics to those who arrive late for a meeting that the heart of any project really lies.

In the B2B / SaaS sector, design is choose the level of abstraction appropriate to solve a business problem. Sometimes the solution is visual, sometimes it is purely logical, and often it is a mixture of both, compressed in time thanks to AI.

The last weeks of 2025 were a perfect laboratory for this thesis. Recently, two consecutive projects in the studio served to validate that design is not a single recipe. Two problems with opposite natures that required us to operate at totally different levels of abstraction.

The first was an in-house tool for a SaaS with a critical urgency of market launch. We decided that the best interface was simply no interface at all. We opted for a command line interface (CLI) to manage the product configuration. This was not a nostalgic retreat, but a strategic decision, sacrificing interface convenience in favour of immediate technical validation. We prioritised architecture and logical robustness over user experience. Operating at this level of abstraction allowed us to validate complex variables with an agility that a traditional UI would have only slowed down. In this scenario, AI accelerated technical feasibility, allowing us to iterate on the product's logic long before we even considered its visual representation.

Days later, the scenario was the reverse, we needed to align a technical and a product team on a new functionality in less than a week. In this case, pragmatism dictated the execution. Instead of designing from scratch, we used AI to assemble already validated patterns and components into a working prototype. If in the previous case we eliminated the UI to unlock the value of the product in the market, here we automated it to accelerate team consensus. The inspiration, as they say, was the delivery date.

However, in this race for efficiency there is a latent danger: the temptation to take shortcuts without a map.

Recently, the CTO of a large company shared with me a concern that resonates throughout the industry. They were detecting a critical knowledge gap in their junior developers, an absolute dependence on the prompt that allowed them to deliver working code, but without understanding the underlying structure or the security or scalability implications of what they were «generating». The problem was not today, but tomorrow: who will be leading and overseeing the systems five years from now if no one learned how to build them from the ground up?

They had opted for a radical solution: forcing the youngest profiles to solve technical challenges by means of software not connected, The only way to do this is through tutorials and physical bibliography. Only through this «forced disconnection» is it guaranteed that the professional will be able to develop the criteria necessary to be more than just a machine operator.

If we delegate execution to AI without understanding the architecture, whether code or design, we lose the ability to correct, to iterate meaningfully, and ultimately to lead. Product design in 2026 is no longer about who has the best command of the tool, but who understands the problem best. AI may give us the answer, but we must remain the masters of the question.

By the way, if anyone needs a strategic boost for their product team or simply wants to brainstorm ideas, our whiteboard is always available.

 

More entries from Design