AI Systems Through a Designer’s Lens: From Brain to Nervous System

AI is becoming a bigger part of the product designer’s toolkit—but understanding what sits behind the interface is becoming just as important as designing the interface itself.

A useful way to understand modern AI systems is to think of them like the human body.

For designers, this analogy is especially powerful because it shifts the conversation from “Which AI tool should I use?” to “What role does AI play in the experience I’m designing?”

1. LLM = The Brain

An LLM is like the brain of an AI system.

It provides reasoning, language generation, interpretation, and the ability to respond to a user’s input.

From a design perspective, this means we can move beyond traditional interfaces where users simply:

Click → Select → Submit → Wait

With LLMs, interfaces can become more conversational and adaptive:

Ask → Explain → Refine → Generate

What this means for designers

Designers need to start thinking about:

How users express intent
How the system interprets ambiguous requests
How AI communicates uncertainty
How users correct AI
How much control the user should retain
How the interface explains why something happened

The challenge isn’t simply designing a chat interface.

It’s designing interaction around intelligence.

2. RAG = Brain + Books

An LLM by itself doesn’t necessarily know your company’s latest information, internal processes, product documentation, or customer data.

That’s where RAG (Retrieval-Augmented Generation) becomes useful.

Think:

Brain + Books

The AI can retrieve relevant information from documents, databases, knowledge bases, or other sources before generating its response.

For designers, this introduces an important UX problem:

Where did this answer come from?

Traditional interfaces usually show a source through a link.

AI experiences can go further.

Designers can create:

Source citations
Document previews
“Used these sources” sections
Confidence indicators
Knowledge freshness indicators
Expandable evidence
Clear distinction between retrieved information and AI-generated interpretation

This creates a new design principle:

Don’t just design the answer. Design the user’s ability to trust the answer.

3. AI Agent = Brain + Hands

This is where things become really interesting for product designers.

An AI Agent doesn’t just generate an answer.

It can take action.

Think:

Brain + Hands

An agent might:

Search information
Create a ticket
Update a database
Send an email
Generate a report
Trigger a workflow
Call APIs
Use other software tools

This fundamentally changes the interaction model.

Instead of:

User → Interface → Action

We start moving toward:

User → Intent → AI → Actions → Outcome

The designer’s role changes

We are no longer only designing screens.

We are designing:

Intent → Planning → Execution → Feedback → Recovery

That means agentic UX needs things like:

Action previews
Approval steps
Progress visibility
Undo / rollback
Permission controls
Error recovery
Human escalation
Audit trails

The key question becomes:

How much autonomy should we give the AI?

Not every action should be fully autonomous.

4. MCP = The Nervous System

If the agent is the hands, something needs to connect those hands to the rest of the digital environment.

This is where MCP (Model Context Protocol) can be thought of as the nervous system.

It helps AI systems connect models and agents with tools, data, and external capabilities in a structured way.

For designers, this is important because it enables AI experiences to move beyond isolated chatbots.

Imagine a design assistant that can understand:

Your product requirements + design system + user research + analytics + codebase + project management tools

Now AI isn’t operating in a vacuum.

It becomes part of the product ecosystem.

What Does This Mean for Product Designers?

The biggest shift isn’t that AI will replace the UI.

It’s that the UI itself is changing.

For years, designers primarily designed:

Screens + Components + Navigation + Workflows

With AI, we’re increasingly designing:

Intent + Context + Intelligence + Actions + Trust

That requires a different mindset.

1. Design for intent, not just interaction

Instead of asking:

“What button should the user click?”

Ask:

“What is the user actually trying to accomplish?”

AI can potentially take care of multiple intermediate steps.

2. Design the AI’s boundaries

An AI system should not feel like it can do everything.

Users need to understand:

What AI can do
What AI cannot do
What AI is currently doing
What requires approval
What happens if something goes wrong

Good AI UX makes these boundaries visible.

3. Design for uncertainty

AI can be confident and still be wrong.

Therefore, confidence, sources, assumptions, and uncertainty become UX elements.

This is one of the biggest differences between traditional software and AI-powered products.

4. Design for collaboration between human and AI

The best AI experiences aren’t necessarily:

AI does everything.

They’re often:

Human decides → AI assists → Human reviews → AI executes

The designer’s job is to find the right balance between automation and control.

The Bigger Design Opportunity

The evolution looks something like this:

LLM
→ AI can think and generate

RAG
→ AI can access relevant knowledge

Agent
→ AI can take action

MCP / Connected Systems
→ AI can interact with an ecosystem

And this creates a completely new design space.

We are moving from designing interfaces people operate to designing systems people collaborate with.

That is a significant shift for UX.

The future product designer won’t just need to understand typography, grids, components, and interaction patterns.

They’ll also need to understand:

AI behavior + context + tools + autonomy + trust + human oversight.

Because when AI becomes part of the product, designing the interface is only one part of designing the experience.

The real question isn’t “Will AI change design?”

It already has.

The more interesting question is:

Are we designing AI as a feature—or designing products around AI as a new interaction model?

I’d love to hear how other designers are approaching this shift. Share your thoughts and experiences in the comments below.

LOADING

Get In Touch