The energy at ViteConf 2025 in Amsterdam was electric. Beyond the expected deep dives into making applications run at lightning speed, something else was brewing in the conference venue. Yes, the Vite ecosystem continues to push boundaries in building blazing-fast tooling and web applications – but beneath all the talk of performance and speed, I noticed a quiet but meaningful shift in how we think about the experience layer of our systems.
We’ve championed User Experience (UX) as the gold standard for years. Then came Developer Experience (DX), recognizing that the people building the tools matter just as much as those using them, a philosophy at the heart of what birthed Vite. Nowadays, a new acronym has entered the conversation, AX (Agent Experience).
This isn’t just another tech buzzword destined to fade into obscurity. AX represents a fundamental evolution in conceptualizing, building, and interacting with software systems in an AI-driven world. It’s the missing piece that bridges human experience and autonomous intelligence, transforming how AI agents interact with our systems, consume our APIs, and collaborate within our software ecosystems.
In this article, we’ll unpack Agent Experience, explore why designing for AI agents is becoming as critical as designing for humans, and discover how this emerging paradigm is reshaping the entire software engineering landscape.
Why experience matters at every layer
Experience isn’t a nice-to-have. It’s the difference between a product that gets adopted and one that gets abandoned. It’s the difference between large initial use and no return users. At every step of the software development lifecycle, experience determines success or failure.
The UX revolution: When users became priority one
Remember the early 2000s? Software worked, sure, but using it could sometimes feel like a punishment. You’d submit a form and get slapped with an error message like “Invalid input in field 7B.” Navigation? Buried three menus deep. And if you couldn’t figure something out, well, there was probably a 200-page PDF manual somewhere.
Then things started to change. Apple and Google showed the world that intuitive design wasn’t just about making things look pretty, it actually gave you an edge over the competition. Suddenly, User Experience became this whole discipline. A philosophy, even. Something you couldn’t just tack on at the end anymore.
Think about a modern e-commerce site. Someone is shopping, adds a few items to their cart, heads to checkout, and then hits a wall. The payment flow is confusing. They can’t find where to edit their shipping address. Their card gets declined, and the error message might as well be in Yoruba language for an English speaker.
Every single one of those friction points? That’s a potentially lost sale. Bad UX isn’t just annoying, it literally costs money.
Good UX does the opposite. It anticipates what people need before they realize they need it. The right path forward should feel obvious. Screw something up? The site helps you fix it without making you feel stupid. It doesn’t overload your brain. And most importantly, it meets people where they actually are, not where we think they should be.
The DX revolution: When developers became users too
Fast-forward to the 2010s. As the software world flourished, developer productivity became the bottleneck. Teams realized developers are users too. The tools they use every day, the frameworks they build with, the APIs they integrate, all of these have an experience layer that directly impacts productivity.
An example is where Vite’s story begins. Before Vite, developers working on large JavaScript applications sometimes faced painfully slow build times. Changing a single line of code meant waiting seconds to see the result. Hot Module Replacement was unreliable. The feedback loop was broken and frustrating.
Vite revolutionized this experience. Suddenly, changes appeared instantly. The dev server started in milliseconds. The experience went from frustrating to delightful. Developer Experience wasn’t just about making things pretty. It was about removing friction from the creative process itself.
Good DX means developers can focus on solving problems, not getting frustrated by their tools. It means clear error messages that point to solutions. It means documentation that answers questions before they’re asked. It means APIs that feel intuitive and frameworks that scale with complexity.
Connecting the dots
Look closely at both revolutions and you’ll see a pattern. When we optimize experience at any layer, we unlock velocity, creativity, and value. UX improvements lead to higher conversion rates and user satisfaction. DX improvements lead to faster development cycles, enjoyable work for developers, and better code quality.
The lesson is clear. Experience is not superficial. It’s fundamental to how systems function, evolve, and scale.
Introducing the Agent Experience (AX)
Now we’re witnessing the next chapter. LLM-based agents are becoming consumers of our systems at scale. They’re calling our APIs, navigating our interfaces, interpreting our documentation, and making decisions based on their understanding of our system’s responses.
There comes a problem.
We’ve been designing systems exclusively for human users and human developers. Agents have different needs, different constraints, and different failure modes. They don’t understand visual interfaces the way humans do. They don’t tolerate ambiguity the same way. They scale differently, fail differently, and succeed differently.
AX is more encompassing, it can be UX plus DX. Agents are used by developers and several others, like non-technical founders, creatives, or product managers. A frustrated agent isn’t just a bad AX, it’s a bad DX and UX.
It’s a distinct discipline that recognizes agents as first-class citizens in our software ecosystems. Just as we once realized that developer experience mattered as much as user experience, we must now recognize that agent experience is equally critical to building systems that work in an AI-driven world.
The question isn’t whether to design for LLM-based agents. They’re already here, consuming our APIs and interacting with our systems. The question is whether we’ll design intentionally for them, or let them stumble through systems built for a human-only world.
Netlify improved AX mentioned at ViteConf 2025
Matt Billman, co-founder and CEO of Netlify, described four critical considerations when implementing AI agents in production to ensure successful deployment.
- Access determines whether agents have the necessary permissions to interact with your product and whether human oversight is required in the workflow.
- Context focuses on providing agents with sufficient knowledge about your product and ensuring they understand how to use it effectively, while also verifying that responses build the appropriate context for users.
- Tools examine whether your product is designed with agent interaction in mind, if you’re providing the right interfaces for agents to work with, and whether you can minimize friction for human users working alongside agents.
- Finally, Orchestration addresses the practical aspects of triggering agent runs from your product, passing context between systems, and providing sandbox environments for safe testing and development.
He illustrated this challenge with Netlify’s CLI, which was originally built for humans. Before the improvement, a human would type netlify deploy into the terminal, hit the Enter button, and follow the prompts. A simple workflow, right? However, this proved problematic for agents. He demonstrated a simulation where Claude Code was asked to deploy to Netlify, showing how the agent repeatedly got stuck in a confusing loop.
The solution was elegantly simple. Netlify added a single instructional line to their CLI that reads: “To create and deploy in one go, use: netlify deploy –create-site <SITE_NAME>”. This clear, actionable guidance provides agents with exactly what they need to complete the deployment in a single command, rather than navigating through multiple interactive prompts. The addition of this one line transformed the agent experience from repeatedly failing in confusing loops to successfully completing deployments. It demonstrates how small, thoughtful changes to developer tools can dramatically improve their usability for AI agents without compromising the human experience.
The Next Wave: What Comes After AX?
If history has taught me anything, it’s that we’re always one paradigm shift away from the next revolution. UX taught us to design for end users. DX taught us to design for builders. AX is teaching us to design for autonomous intelligence.
So what comes next?
The truth is: I don’t know. We’re writing that story together, one system, one API, one experience at a time.
What I do know is that the best time to think about the next paradigmx is before it arrives. By the time everyone agrees on the name, the pioneers have already moved on to solving the next set of problems.
Hence, the real question is, will you be ready when it arrives?
