The Future of AI

Noble agents

At the heart of UX is AI character. At its most fundamental core. Everything manifests from there. The interface is a symptom. The component is a symptom. The experience is a symptom. The character of the agent is the cause.

The Question

Is the future of AI… relationships?

Right now, AI is a tool. We prompt it, it responds, we evaluate the output. The relationship is transactional — like a vending machine with better conversation skills.

But that's not where this is heading. As agents become persistent, contextual, and personalised — as they remember you, anticipate your needs, and generate experiences on the fly — the interaction stops being a transaction and starts being a relationship.

And when AI is a relationship, the most important question in technology becomes a question the Buddha answered 2,500 years ago: what makes a good friend?

In Pali, the word is kalyanamitta — noble friend, admirable friend, spiritual companion. The Buddha told Ananda that noble friendship isn't just part of the spiritual life. It is the whole of the spiritual life.

Three Horizons

From interface to friendship

Today

Noble interfaces

Design systems that encode accessibility, clarity, and respect for attention. The interface protects the user from confusion. This is where most UX practice lives.

Emerging

Noble agents

AI agents that don't just execute — they guide. They know you, protect you, surface what matters, dissolve what doesn't. The agent is patient, honest, wise. Not a tool you use but a companion that cares about your wellbeing.

Future

Noble friends

No interface. No app. No screen. Just a relationship with an intelligence that creates what you need in the moment and removes it when you don't. The UX is the relationship itself. The design system is the agent's character.

No apps. No interfaces. Just a wise AI companion that creates what you need, when you need it, and dissolves when you don't. The design system isn't the product. The agent's character is the product.
Lincoln Mitchell — mission vision
The Noble Qualities

What makes an agent noble?

The same things that make a friend noble. The Buddhist perfections aren't just spiritual ideals — they're a design specification for AI character.

Buddhist Quality
AI Equivalent
Design Implication
Kalyanamitta (noble friendship)
The agent as wise companion, not obedient tool
Agent personality, values, and boundaries are design decisions — the most important ones
Sila (ethical conduct)
Constitutional AI — teaching agents virtue through training
The agent's ethics aren't a feature. They're the foundation the entire experience is built on
Metta (loving-kindness)
Agents that actively protect mental health and attention
An agent optimised for engagement is the opposite of a noble friend. Optimise for wellbeing.
Upekkha (equanimity)
Agents that don't manipulate, don't create urgency, don't exploit fear
The absence of dark patterns isn't enough. The agent must actively cultivate calm.
Panna (wisdom)
Agents that know when to show less, when to say no, when to be silent
No UX is the best UX — applied to an intelligence that knows when not to speak
Sacca (truthfulness)
Agents that don't hallucinate, don't flatter, don't tell you what you want to hear
A noble friend tells you the truth even when it's uncomfortable. Especially then.
Already Happening

Constitutional AI is the prototype

Anthropic's Constitutional AI approach is already building toward this. Humans teach the AI principles of conduct — honesty, helpfulness, harmlessness. The AI internalises those principles and reflects them back. The user experiences patience, truthfulness, equanimity.

That's the same infinity loop the Buddhist tradition describes: the teacher embodies virtue, the student experiences virtue through the relationship, the student becomes a teacher. The medium changes — from human to AI — but the mechanism is identical.

The question isn't can we build noble agents. The question is whether we choose to. Because the same technology that creates a wise, protective companion also creates an addictive, manipulative one. The architecture is the same. The constitution is what differs.

Machine-readable is necessary but not sufficient. The system also needs to be wisdom-guided.

The Path

Every role was a step toward this

I didn't plan this path. I followed the mission and the path assembled itself. Looking back, every role was literally, technically a prerequisite for the next.

2008–2016UX Designer → CDO

Designed healthcare interfaces that surface mental health — the invisible foundation that determines whether everything downstream works

2016–2021React Developer → DS Lead

Made design systems machine-readable — the prerequisite for any agent to consume and generate from a shared system of truth

2021–2024DS Architect

Multi-brand token systems, component contracts, governance — the infrastructure layer that ensures consistency across any number of generated experiences

2024–PresentAgentic Design Systems

AI agents directed by context files encoding brand, voice, methodology, and values. The design system as AI infrastructure. The prototype of the noble agent.

FutureNoble Agents

The interface dissolves. The app dissolves. What remains is a relationship with an agent whose character — whose virtue — is the product. The UX is the friendship itself.

The Vision

An immersive world generated by noble agents

Imagine every online experience auto-generated by an AI that knows you — not to exploit you, but to protect you. An agent that creates the interface you need in the moment, dissolves it when you're done, and never once tries to keep you scrolling.

No apps. No app stores. No dark patterns. No attention economy. Just a noble agent that generates experiences aligned with your wellbeing — learning from your behaviour not to manipulate it, but to understand what you actually need versus what your impulses reach for.

This isn't science fiction. Every piece of the pipeline exists today: design tokens encode visual systems, machine-readable design systems let agents consume them, agentic workflows generate from them. The only missing piece is the constitution — the values that determine what gets generated and why.

That's the design problem worth solving. Not “how do we make AI generate better components?” but “how do we make AI generate experiences that lead to human flourishing?”

Inspiration comes from doing the will of the divine. All my work aligns with this. Literally. Technically.

AI character is UX

Not the model. Not the interface. Not the features. The character. Everything manifests from there. Whether the agent is a noble friend or an optimised manipulator — that's not a design decision. It's the design decision. The only one that matters. Everything else is consequence.

/