Matias Gonzalez

Helping founders drowning in Product chaos I cut through the noise and help you move Strategy | UX | Fast action

[Matias Gonzalez @matigon]

Figma to Code... Ok, sure...

Code to Figma... wait, what?

https://t.co/qAMM7PoCfa

#ProductDesign #prototype #figma

[Matias Gonzalez @matigon]

I see constraints are now closer to decision making, judgement and taste.
But my mind goes to... how can designers re-shape their positioning to create leverage to be able to design instead of fix AI generated pieces?

[Matias Gonzalez @matigon]

I’m looking for Product or Graphic Designers in Madrid. (Europe works too.)

I want to sit down (coffee counts) and talk about what design actually looks like in 2026.

If you feel this too, DM me.

[Matias Gonzalez @matigon]

I feel that, changes are being so deep that is the mindset what's shifting...

I see some designers pushing for things to stay the same (sticking to tools or processes created for constraints that are fading away).

Some others are adopting every new tool and surfing the wave.

And clients (entrepreneurs/PMs/POs) are crafting tasks that use to be executed by designers.

These things are amazing from a productivity and speed standpoint. What I wonder is if they are hurting what Design as a craft can bring to the world.

[Matias Gonzalez @matigon]

I've been feeling a strong pull towards analog physical design lately.

I wonder if I should start creating and post my way journey on "going against the grain" while using AI tools to get there...

[Matias Gonzalez @matigon]

Taste and intuition take an increasingly important role for modern designers...

[Matias Gonzalez @matigon]

Moving fast without a map just gets you lost faster.

Speed without discovery is just fast guessing.
👉 https://t.co/6EXcZFdvpd

[Matias Gonzalez @matigon]

Product design has shifted.

Discovery must work harder, not disappear, to protect users and products.

[Matias Gonzalez @matigon]

Most companies involve design too late because they don’t see it as a business lever.

There’s a low-cost, repeatable way to make design matter in decision-making.

I talk about it here ↓
https://t.co/B7NNOZd5V4

[Matias Gonzalez @matigon]

Multitasking isn’t a skill.
It’s a tax on every important thing you’re trying to build.
So if you want compounding results, stop scattering your effort. Protect 2–3 hour blocks.

Do the one thing that actually moves the business.
https://t.co/g1EMwlknAU

[Matias Gonzalez @matigon]

Teams don’t lose sight of the user because they’re inexperienced.
They lose sight because they think they know them.

And the business pays for it.

https://t.co/Kj15A3LY0i

#Strategy #design

[Matias Gonzalez @matigon]

Sometimes the thing you love doing the most doesn’t end up being what defines your creative story.

From time to time, beauty can be something that simply disrupts the mainstream. Something that shifts the meaning of what “beautiful” (or “creative”) means.

Mondrian used to paint trees, but his creative exploration led him to abstract forms. The latter ended up defining his impact on the art world.

It made me think there are things you just can’t plan for. Follow my flow, enjoy the ride, let luck do its thing.

Here’s a UI palette inspired by his work.

[Matias Gonzalez @matigon]

I'm starting a series to explore the beauty that came before us: #palettestudies

"Art was barely ever appreciated in its own time."

Was it? Is it today? How can I tell when something is beautiful?

Beauty... such an abstract thing.

But you know it when you see it, right? Or does it have to belong to an older time, loaded with suffering and abstract meaning, to be appreciated?

I find modern visual design often achieves "being perceived as beautiful" by commoditizing replicable trends that yearly deprecate the old ones and bring a new version of beauty.

And, more often than not, like sheep craving the love of their pastor, we follow.

So, I’ll start exploring what came before us and bring it into today’s function.

Today, I pulled this palette from one of Vincent’s self-portraits. His chaotic paintings helped me understand that mainstream is not always the way.

[Matias Gonzalez @matigon]

The old “upload an image of your receipt” might be over… hoy do you approach this issue?

Some say companies are already looking for blockchain solutions to address this issue.

[Deedy @deedydas]

You can use 4o to generate fake receipts.

There are too many real world verification flows that rely on “real images” as proof. That era is over.

[Matias Gonzalez @matigon]

Although my design flow execution is now heavily AI oriented, the meaningful substance of what I'm creating doesn't come from AI at all.

A strong foundation is what supports a building, no matter the construction materials.

[Matias Gonzalez @matigon]

Here's my quick framework for auditing UX these days:
1. Can a user predict what will happen before clicking?
2. When AI fails, is next action obvious?
3. Do you show confidence scores or hide when AI is guessing?
4. Is there a clear fallback when AI has low confidence?
5. Can users see/edit the context window feeding the AI?

Fail any of these = you have an AI UX problem.

Which one does your product Fails or Nails?

[Matias Gonzalez @matigon]

The best AI UX isn't the one that looks pretty.
It's the one that fails gracefully when the model hallucinates.

Most designers don't even know what a confusion matrix is.

That's why your AI features have a 10% retention rate.

[Matias Gonzalez @matigon]

You can't spit a Chatbot in your homepage and expect your product interaction problems to be over.

[Matias Gonzalez @matigon]

User experience might fall strongly as Vibe coding combines the "Viber's" (how will we call devs that don't code?) output anxiety with poor AI execution.

Remember, your user deserves better...

[Benjamin Todd @ben_j_todd]

Why can AIs code for 1h but not 10h?

A simple explanation: if there's a 10% chance of error per 10min step (say), the success rate is:

1h: 53%
4h: 8%
10h: 0.002%

@tobyordoxford has tested this 'constant error rate' theory and shown it's a good fit for the data

chance of success declines exponentially

[Matias Gonzalez @matigon]

This is a great way of building! One specific use-case with a clear goal. Kudos!

I’d try to identify what active users are finding most valuable (since people often prefer different ways of logging emotions), and lean into that signal.

Have you had the chance to run any user interviews yet?