Foundations of AI Communication
When we think about digital design, we usually picture screens, structures, components, and visual systems. But in this new era we’re living in, designing also means learning to communicate with tools that help us create, not with technical commands, but with natural language. And that’s where a new skill comes into play: knowing how to “talk” to artificial intelligence.
Tools like Lovable, V0, Cursor, or Claude don’t work like classic assistants. It’s not enough to give them an order. They work best when we explain what we want, with intention, context, and clarity. They’re not machines that execute without thinking; they’re collaborators that understand, as long as we express ourselves well.
That changes everything: design stops being a one-way process and becomes a dialogue. And as in any conversation, what you get depends on what you say, how you say it, and how you react to the response. Knowing how to ask for what we want, structuring our ideas, iterating, and avoiding the most common mistakes is now part of the design job
From giving orders to having a conversation
One of the biggest shifts in working with AI is changing how we address it. We’re no longer talking to a machine that just executes commands. We’re talking to a tool that interprets, suggests, and completes based on what we tell it and how we tell it.
Rigid instructions, as if we were writing code or giving orders to a mindless assistant, no longer work as before. Today, tools like Lovable, V0, or Cursor understand natural language. That means we can (and should) speak to them as if we were explaining an idea to a teammate.
It’s not about being technical; it’s about being clear. The key is to convey intention, not just action.
It’s not the same to say: “Make me a website to sell bicycles”
as to say:
“I need an online store focused on urban bicycles, aimed at people living in cities who use cycling as a sustainable alternative to traditional transport. The main focus will be on electric models, so that category must stand out on the home page.
For the design, I’m looking for a clean, modern style with plenty of white space, a sans-serif typeface, and a visual aesthetic that conveys movement, flow, and lightness. I want large, high-quality images, especially on the product page, which should also include filters for type of use (daily, sport, folding, electric, etc.).”
Both statements might seem similar in intention, but they’re not. The first is a generic order with no detail on expectations. It could apply to a sports shop, a general marketplace, or even a corporate site. Without enough information, the AI will fill the gaps on its own — and it’s unlikely to get it right.
The second option works like a real conversation with a teammate: it sets the usage context, target audience, product focus, desired aesthetic, minimal structure, and commercial intent. It gives the AI the necessary coordinates to build a first proposal consistent with our vision. It’s not about overwhelming the AI with infinite instructions, but about giving it direction.
The key isn’t talking more, but talking well. If you’d give references, examples, and nuances to a person so they understand your idea, you should do the same with AI. That’s how it becomes truly useful. Otherwise, you’ll just get a generic version — soulless and unfocused.
The value of context
When we communicate with AI, we’re not talking to someone who already knows our project, brand, or goals. Each time we start a conversation, it begins from zero. That’s why the context we provide is never excessive; it’s the foundation on which everything that follows will be built.
Imagine explaining an idea to a new team member. You’d tell them what you need, why you need it, who it’s for, and what you have in mind visually. With AI, the principle is the same: the clearer those pieces are, the more accurate the response will be.
But giving context doesn’t mean writing an endless text. It means organizing the information so the AI can make the right decisions. Here are some elements to include from the start:
- Goal: What do you want to achieve? What’s the main function of the app, site, text, or component? What problem does it solve?
- Audience: Who will use this? What do they know, need, or how do they behave? Designing for experts isn’t the same as for first-time users.
- Style: What visual, narrative, or emotional tone do you expect? Are there brands or products that serve as references? What feelings do you want to convey, and which should you avoid?
- Constraints: What must not be included? What conditions must be respected? This can cover technical, legal, accessibility, branding issues, or just personal preference.
Common mistakes and how to avoid them
Starting with AI tools can be as exciting as it is bewildering. At first, many people feel that “the AI didn’t understand what I wanted” or that “the result looks nothing like what I imagined.” That can lead to frustration or, worse, abandoning a tool that, used well, can be a great ally.
The reality is that most stumbles aren’t the AI’s fault, nor the user’s, but stem from the communication between them.
When we treat AI like a black box that should spit out brilliance just because we say “make me a nice app,” we forget that what really makes the difference isn’t the tool itself, but how we express ourselves with it.
These mistakes aren’t signs you’re doing things wrong — they’re part of learning. Spotting them early and adjusting your communication will get you better answers, reduce unnecessary iterations, and make you feel more comfortable and in control.
Some of the most common errors are:
- Vague instructions
- Stuffing too much into a single prompt
- Not giving enough context
- Failing to define success criteria
- Skipping iteration
- Overloading references without explanation
- Ignoring limits and constraints
- Expecting instant magic
- Confusing inspiration with copying
If you want to bring this way of working to your team, the Vibe Coding for Designers Specialization Program at UX Learn, our training platform, covers all these concepts in depth. You’ll learn to design through language, think about product through conversation, and use vibe-coding tools to go from idea to prototype without technical barriers.
NOTE: This training is only available in Spanish.
This is a translation of the following article from our corporate website:
