In this interview, e27 speaks with Paruey Anadirekkul, Founder & CEO of Spacely AI, a generative AI platform focused on spatial design. As AI continues expanding beyond text and image generation into specialised professional domains, Spacely AI explores how automation can support architects, interior designers, and real estate professionals in visualising ideas faster and making earlier, more confident decisions.

This conversation forms part of e27’s broader AI Pulse coverage, which examines how organisations across the region are building, deploying, and governing AI in real-world settings.

Generative AI for spatial design workflows

e27: Briefly describe what your organization does, and where AI plays a meaningful role in your work or offering.

Paruey: Spacely AI is a generative AI platform built for spatial design — helping professionals turn ideas, floor plans, and rough concepts into realistic 3D visualizations and design iterations in minutes. Our customers are primarily interior designers, architects, and real estate professionals, with a large portion based in the US and Europe.

AI plays a central role in both our product and internal operations. On the product side, we use proprietary generative models and 3D algorithms to automate rendering, layout variations, lighting adjustments, and even 2D-to-3D model conversion. This shortens early-stage design cycles dramatically, where most cost and timeline decisions are made.

Internally, AI supports product development, customer success, marketing experimentation, and data analysis. We treat AI not as a feature add-on, but as infrastructure — embedded into workflows to reduce repetitive tasks and allow our team and customers to focus on higher-value creative and strategic decisions.

Reducing friction in design and property decisions

e27: What is one concrete way AI is currently creating value within your organisation or for your users or customers?

Paruey: One clear way AI creates value is by reducing friction at the decision stage.

For real estate brokers and agencies use Spacely AI to help buyers see the true potential of a property. Instead of relying on imagination when viewing an empty or outdated space, buyers can instantly visualize renovations or new layouts. This reduces hesitation, shortens the purchase decision cycle, and increases conversion because clients can see what they are buying — not just what exists today.

For Interior Design Company, the value is in faster client alignment. Early in a project, most delays come from back-and-forth discussions about style, mood, and direction. By generating multiple realistic design options quickly, Spacely AI helps clients react to something concrete. This shortens alignment time, reduces revisions, and allows the team to move into detailed design work faster.

In both cases, the outcome is not just speed. It is clearer communication, stronger confidence in decisions, and better commercial results.

Also read: AI Pulse Exclusive: How CoBALT is designing AI that teams can actually trust

Balancing model performance and economics

e27: What was a key decision or trade-off you had to make when adopting, building, or scaling AI?

Paruey: One key trade-off we faced was between model performance and economic sustainability.

Early on, we realized that generic models did not perform well for our target users. Interior designers and architects prompt very differently from casual users. So we invested time in fine-tuning and optimizing models specifically for spatial design workflows. That improved output quality and consistency, but it also increased compute costs.

At the same time, we had to balance cost, speed, and quality for different user tasks. High-fidelity rendering requires more GPU resources, while quick concept iterations can run on lighter infrastructure. Choosing when to use which model became a product decision, not just a technical one.

Gross margin is critical in AI SaaS. We had to design our pricing and token model carefully to protect margins while still delivering meaningful value to customers. The lesson was that AI capability alone is not enough — the real challenge is building a system where performance, user experience, and unit economics all work together sustainably.

Adoption momentum and integration challenges

e27: Looking back, what has worked better than expected, and what proved more challenging than anticipated?

Paruey: What worked better than expected was adoption speed. Once designers saw that AI could generate realistic concepts in minutes, not days, the willingness to experiment was high. Many professionals who were initially skeptical became regular users after seeing practical results in client meetings.

What proved more challenging was workflow integration. AI can produce impressive outputs, but using it effectively requires learning how to prompt well, iterate, and interpret results. There is still a skill curve. The value comes not from one-click magic, but from knowing how to guide the system.

Another challenge is the pace of AI improvement. Our rendering accuracy today is significantly better than it was six months ago. Costs, speed, and quality have improved rapidly. However, user perception often lags behind. Some customers still remember early limitations — like distorted elements — even though those issues have been resolved. Managing expectations in a fast-moving technology landscape has been just as important as improving the models themselves.

Rethinking workflows for AI adoption

e27: What is one lesson about applying AI in real-world settings that leaders or founders often underestimate?

Paruey: One lesson leaders often underestimate is that AI adoption is not a plug-and-play purchase.

Buying access to an AI tool does not automatically produce better outcomes. The real value only appears when teams rethink their workflows around it. Many organizations try to layer AI on top of existing processes without changing how work is structured. That usually leads to frustration or underuse.

In our experience, the biggest gains come when teams revisit where time is spent, where decisions are delayed, and where iteration cycles are slow. Then AI can be embedded intentionally into those pressure points. Adoption requires training, experimentation, and sometimes redefining roles — not just software procurement. AI is most powerful when paired with operational redesign, not treated as a shortcut.

Also read: AI Pulse Exclusive: How Asia AI Association is advancing human-centred AI across the region

Starting with outcomes, not models

e27: Based on your experience, what is one practical recommendation you would give to organisations that are just starting to explore or scale AI?

Paruey: One practical recommendation is: don’t start with the technology.

AI models change every few months — in cost, speed, and capability. If you start by choosing a model, you risk building around something that may soon be obsolete. Instead, start with outcomes. Define what you want to improve, map the current workflow, and identify the specific bottlenecks or repetitive tasks that can be automated or augmented.

Only after that should you evaluate which AI model fits the job today. Treat AI as a modular component. Design your architecture and processes so you can swap models as the landscape evolves. The advantage doesn’t come from picking the “best” model — it comes from building flexible workflows that can continuously improve as AI advances.

AI becoming invisible infrastructure

e27: Over the next 12 months, how do you expect your organisation’s use of AI, or the role of AI in your industry, to evolve?

Paruey: Over the next 12 months, AI will become less visible and more expected.

In our industry, AI will shift from being a standalone tool to something embedded directly inside design workflows. Users won’t “use AI” as a separate step — it will be integrated into rendering, planning, cost estimation, and collaboration processes. The focus will move from generating impressive outputs to improving measurable outcomes like faster alignment, reduced rework, and higher win rates.

We also expect leadership expectations to mature. Instead of being impressed by what AI can generate, leaders will ask: did this reduce costs, shorten timelines, or increase revenue? The conversation will move from capability to accountability. AI will become operational infrastructure — evaluated on business impact, not novelty.

Adaptability over technical advantage

e27: Anything else you want to share with the audience?

Paruey: One final thought: AI will not reward the most technical companies — it will reward the most adaptive ones.

The advantage won’t come from having access to the latest model, because everyone does. It will come from how quickly teams experiment, measure impact, and adjust workflows. The organizations that win will treat AI as an ongoing capability, not a one-time transformation project.

Also, we should be honest: AI is not perfect. It makes mistakes. It requires oversight. But so do humans. The real opportunity is designing systems where human judgment and AI speed complement each other. When that balance is right, the results are not just faster — they are better decisions made earlier, where they matter most.

Also read: AI Pulse Exclusive: How Explico is building AI teachers can actually rely on

Designing AI for practical creative workflows

This conversation highlights how AI is increasingly moving into specialised professional domains beyond general content generation. In areas like spatial design, real estate, and architecture, the emphasis is shifting toward faster iteration, clearer decision-making, and integrating AI directly into everyday workflows. As adoption matures, organisations may find that the real advantage lies less in the models themselves and more in how effectively teams adapt processes to work alongside AI.

For more interviews, analysis, and real-world perspectives on how organisations across the region are applying AI in practice, click here.

—

Enjoyed this read? Don’t miss out on the next insight. Join our WhatsApp channel for real-time drops.

The e27 team produced this article

We can share your story at e27 too! Engage the Southeast Asian tech ecosystem by bringing your story to the world. You can reach out to us here to get started.

Featured Image Credit: Spacely AI

The post AI Pulse Exclusive: How Spacely AI is bringing generative AI into spatial design workflows appeared first on e27.