Our Learnings
Much of our work lives behind NDAs.
So instead of traditional case study imagery, we’ll show you a pretty video.
It’s a visualisation of the codebase of the project evolving over time. Files appear, branches grow, connections form, things evolve. You’re watching a project slowly take shape.
What we can share is the thinking behind some of our projects. The questions we explored, the assumptions we tested, and the principles that emerged along the way.
What follows is a series of those explorations, shared as honestly as we can.
Project: An Adaptive Brand Experience
Testing how adaptive systems change the way people explore and engage with brands.
The future of brand experiences isn’t immersive. It’s adaptive.
Immersion on its own doesn’t mean much anymore. You can surround someone with beautiful visuals and still lose them in seconds. What matters is whether an experience adapts. To curiosity. To hesitation. To intent. As conversational AI starts to sit at the centre of spatial experiences, we’re learning that the real craft isn’t world-building. It’s designing behaviour. When to lead, when to listen, and when to get out of the way entirely.
For a long time, digital experiences have been designed around certainty. Fixed journeys. Predefined outcomes. Clear success metrics. That made sense when interaction was limited to screens and clicks. But conversational systems don’t behave like interfaces. They respond. They infer. They interrupt. And spatial environments remove the final safety net. There’s no obvious “next” button floating in mid-air. Once you put those two things together, you’re no longer designing a journey. You’re designing a relationship. And relationships only work when systems are allowed to adapt in real time.
That’s why we’ve been using prototypes as behavioural sandboxes rather than previews of finished work. Not to show what’s possible, but to understand what happens when control is loosened. In one recent build, we deliberately stripped away menus, linear narratives and obvious calls to action. Instead, we focused on a conversational layer that could listen, respond and step back when needed. The goal wasn’t polish. It was signal. Where people paused. What they asked unprompted. When the system should lean in, and when silence was the better response.
What surprised us most was how little guidance people actually wanted. When given the freedom to explore, most didn’t get lost or frustrated. They became more intentional. Conversations slowed down. Questions got more specific. Instead of asking broad, brand-level things, people gravitated toward details that matched their own values and interests. Just as importantly, they ignored huge parts of the experience entirely. Not because it failed, but because relevance is personal. That moment. When people confidently skip what doesn’t matter to them. Is something traditional experience design rarely allows for, but adaptive systems make possible by default.
Not everything worked. Our early instinct was to be helpful too quickly. The system jumped in with context and suggestions before people had fully formed what they were curious about. On paper, it felt elegant. In practice, it subtly broke the spell. Even a well-timed prompt can feel intrusive if it arrives before intent is clear. We learned that adaptability isn’t about responding fast. It’s about responding at the right moment. Sometimes the most intelligent thing a system can do is wait.
That single failure ended up shaping how we now think about adaptive experience design.
Principle 1. Timing beats speed.
One of the biggest myths around conversational systems is that faster is always better. It isn’t. What matters is whether a response arrives at the moment someone is ready to receive it. In adaptive experiences, timing is the difference between something feeling attentive and something feeling interruptive. A system that speaks too soon takes control away. A system that waits earns trust. Designing for timing means reading hesitation, silence and curiosity just as carefully as words. It means accepting that responsiveness isn’t about immediacy. It’s about restraint.
Principle 2. Agency beats guidance.
Traditional experience design is built around helping people move forward. Adaptive experiences work best when they let people choose where forward even is. When we removed predefined paths, something interesting happened. People didn’t freeze. They explored. They followed curiosity instead of instruction. Guidance still mattered, but only when it was invited. The role of the system shifted from narrator to companion. From telling people what mattered to responding to what already did. That shift. From direction to permission. Is where adaptive experiences start to feel genuinely human.
Principle 3. Behaviour beats spectacle.
It’s easy to mistake immersion for impact. Big visuals. Rich worlds. Impressive tech. But none of that matters if the experience doesn’t change how someone behaves within it. Adaptive systems force a different question. Not “what did they see?”. But “what did they do next?”. Where did they linger. What did they ignore. What did they ask for without being prompted. Those signals are far more valuable than completion rates or dwell time alone. When you design for behaviour, spectacle becomes a tool. Not the goal.
This shift isn’t about immersive tech, conversational AI, or whatever the next platform happens to be. It’s about a deeper change in how people expect systems to behave around them. As interfaces fade, experiences stop being something you move through and start becoming something you negotiate with. Brands that cling to control will keep designing impressive moments that people forget. The ones that invest in adaptability will build experiences that listen, respond and evolve. That’s the real opportunity. Not to make things louder or bigger. But to make them more human.
At Astral City we build brand experiences that listen before they lead.
Project: Decision-Making Product Companion
Exploring how conversational products can build confidence before offering answers.
The hardest part of product design isn’t personalisation. It’s confidence.
Most product problems don’t start with a lack of options.
They start with too many.
People arrive with intent, but it’s usually fuzzy. A feeling. A direction. A vague sense of what they want to change. What stops them moving forward isn’t capability or tooling. It’s uncertainty. Am I heading in the right direction. Is this a good choice. What happens if I commit.
A lot of AI-powered products talk about personalisation as the solution. But personalisation on its own doesn’t help people decide. It just narrows the menu. The real challenge is helping someone build enough confidence to take the next step without feeling rushed, judged or overwhelmed.
That tension became the centre of a recent product exploration we worked on. The brief wasn’t to automate decisions or “design the perfect outcome”. It was to understand how a conversational system might help people think more clearly about what they actually want, while staying grounded in real-world constraints.
So we treated the product less like a tool and more like a thinking partner.
Instead of jumping straight to recommendations, the experience focused on sense-making. Helping people articulate taste, priorities and trade-offs in their own words. Letting preferences emerge through conversation, iteration and small reactions rather than forcing everything into a form or preset. The system could suggest, visualise and organise, but it never assumed it knew best.
What we were really testing was a simple question.
At what point does assistance become pressure.
One thing became obvious very quickly. When the system moved too fast, people disengaged. Even helpful suggestions felt premature if they arrived before someone had fully externalised their thinking. On paper, those moments looked efficient. In practice, they short-circuited confidence.
It sounds obvious now. It wasn’t at the time.
We learned that progress doesn’t always look like forward motion. Sometimes it looks like reflection. Sometimes it looks like removing options rather than adding them. Sometimes the most useful thing a system can do is help someone understand why they’re hesitating, not push them past it.
That led us to a set of product principles we now come back to often.
Principle 1. Confidence beats speed.
Fast isn’t always helpful. Especially when people are forming opinions, not executing tasks. A product that rushes to an answer can accidentally take ownership of the decision, leaving the user less certain, not more. Designing for confidence means allowing pauses, backtracking and uncertainty to exist without penalty. Momentum comes from clarity, not velocity.
Principle 2. Structure beats suggestion.
Recommendations are seductive. They feel like value. But without structure, they just add noise. The most effective moments came from helping people organise their thinking. Grouping ideas. Surfacing patterns. Making trade-offs visible. Once structure was in place, suggestions felt welcome. Without it, they felt intrusive.
Principle 3. Help people decide. Don’t decide for them.
The goal was never to produce a “perfect” outcome. It was to help people feel ownership over the one they chose. That meant the system had to be comfortable stepping back. Letting users override it. Letting them change their mind. Letting them ignore entire branches of exploration without consequence. Confidence grows when people feel in control, not when they’re optimised.
What this reinforced for us is that good product design in an AI-heavy world isn’t about intelligence. It’s about judgment. Knowing when to surface insight. When to hold context. When to stay quiet. And when to let a human sit with uncertainty long enough to resolve it themselves.
This isn’t just relevant to conversational AI or spatial products. It applies to any system that claims to help people make decisions. As products become more proactive, the risk isn’t that they’ll do too little. It’s that they’ll do too much, too soon.
The opportunity isn’t to remove friction entirely.
It’s to remove the right friction, at the right moment.
At Astral City we build products that know when to guide, and when to wait.