Right now, we treat the interaction between a person and software as an interface. But what about software built for interaction with an AI, another intelligence? That is a whole new Pandora’s box.

Two parallel dimensions are showing up: human-to-interface and agent-to-agent. Designers are now expected to think and design for both at the same time.

Does this approach even have a chance to survive? I think no.

Because what is the point of building an interface and a backend “interface” for two different types of use? It is inefficient for everyone.

Why?

• There is still some kind of interface where a human gives tasks to their agent. ↓

• The agent goes online, gathers information, and completes the task. But it is analyzing information that was created for humans, so the result is lower quality.

• The agent does not care what style your website has.

• The agent does not care if you followed accessibility standards.

• The agent does not understand or fall for manipulative slogans, and it has no emotions to trigger those old marketing-driven impulse purchases.

• The agent does not care about brand, positioning, and all the rest.

In practice, the agent serves as a high-quality filter for now.

So why is a designer writing this?

Because the current industry demand is “designer-developer.” But why? Is it worth investing time in learning development and becoming a developer? I do not see the point. If everything is moving toward code being written by AI, why develop a skill that is already fading?

What do you get from a designer who codes? You just save on the frontend part, which will eventually fade out anyway under the agent-to-agent approach.

Leave agents to developers.
↓
Focus on data visualization instead of dumb interfaces. People will not read your text anymore. They will read, or listen to, a version rewritten by their agent. And the only thing the agent can reliably pull from you is the kind of visual information it cannot reformat away. That is what you should focus on.

For now, what stays constant are the physiological channels humans use to consume information: sight and hearing. I am ignoring touch, smell, and taste, even though that is probably just a matter of time.

The world is overloaded with information, so
people learned to stop reading and start scanning.
And if the image attached to a post is not interesting, nobody reads the text. That is why most people are visual now.

I predict that audio as a channel of interaction with an agent will be mostly for commands. But for consuming information, it will be mostly visual.

So, dear designers, if you already started learning code, learn it in the domain of complex data visualization.

If this essay helped you, a tip is appreciated. It helps me keep writing visual pieces like this.

Right now, we treat the interaction between a person and software as an interface. But what about software built for interaction with an AI, another intelligence? That is a whole new Pandora’s box.

Two parallel dimensions are showing up: human-to-interface and agent-to-agent. Designers are now expected to think and design for both at the same time.

Does this approach even have a chance to survive? I think no.

Because what is the point of building an interface and a backend “interface” for two different types of use? It is inefficient for everyone.

Why?

• There is still some kind of interface where a human gives tasks to their agent. ↓

• The agent goes online, gathers information, and completes the task. But it is analyzing information that was created for humans, so the result is lower quality.

• The agent does not care what style your website has.

• The agent does not care if you followed accessibility standards.

• The agent does not understand or fall for manipulative slogans, and it has no emotions to trigger those old marketing-driven impulse purchases.

• The agent does not care about brand, positioning, and all the rest.

In practice, the agent serves as a high-quality filter for now.

So why is a designer writing this?

Because the current industry demand is “designer-developer.” But why? Is it worth investing time in learning development and becoming a developer? I do not see the point. If everything is moving toward code being written by AI, why develop a skill that is already fading?

What do you get from a designer who codes? You just save on the frontend part, which will eventually fade out anyway under the agent-to-agent approach.

Leave agents to developers.
↓
Focus on data visualization instead of dumb interfaces. People will not read your text anymore. They will read, or listen to, a version rewritten by their agent. And the only thing the agent can reliably pull from you is the kind of visual information it cannot reformat away. That is what you should focus on.

For now, what stays constant are the physiological channels humans use to consume information: sight and hearing. I am ignoring touch, smell, and taste, even though that is probably just a matter of time.

The world is overloaded with information, so
people learned to stop reading and start scanning.
And if the image attached to a post is not interesting, nobody reads the text. That is why most people are visual now.

I predict that audio as a channel of interaction with an agent will be mostly for commands. But for consuming information, it will be mostly visual.

So, dear designers, if you already started learning code, learn it in the domain of complex data visualization.

If this essay helped you, a tip is appreciated. It helps me keep writing visual pieces like this.