A
Architizer Editors
Guest
- Thread Author
- #1
Architects: Want to have your project featured? Showcase your work by uploading projects to Architizer and sign up for our inspirational newsletters.
Portugal needs more housing, but delivering it without compromising architectural quality is becoming increasingly difficult. Saraiva + Associados (S+A), a Lisbon-based architecture practice with 30 years of experience, has spent the past two years trying to close that gap using AI.
With more than 100 people on staff and operating across nine countries, S+A is the largest architecture-only practice of its kind in Portugal. Like many practices around the world, S+A began experimenting with AI out of curiosity. Now, the firm is integrating AI into the architectural decision-making process and helping clients understand the consequences of design choices before they become expensive compromises.
At first, the office ran a wider evaluation across multiple AI visualization tools and kept hitting the same issue: geometry drift. “The first tests we ran were crazy,” says Luís Barros, Chief Operating Officer and Architect at S+A. “A designer had a concept for a house, and by the end it had become something completely wild and different.” The tools were good at generating images, but they were bad at staying inside the idea.
“For us, it’s never about sacrificing design to reduce costs. It’s about designing within a realistic framework that balances architectural quality, budget and delivery time,” says Tiago Tavares, Senior Partner and Architect at S+A. If an AI tool changes the architectural concept simply to produce a visually attractive image, the rendering is not useful anymore.
Saraiva + Associados (S+A) architects Manuel Banazol, Luis Barros and Tiago Tavares
The S+A team didn’t give up and decided to set up an internal investigation group with twelve people testing AI tools against the firm’s actual workflow. One tester brought Visoid to the table, a tool that, unlike the others, held the line on geometry, followed the concept and rendered it. “We’re so careful about following our clients’ requests closely, it was easy to see that a tool like this made perfect sense for us,” says Barros.
Visoid is a rendering tool built specifically for architects. Using AI, it turns design models into high-quality visuals in minutes without the usual rendering workflows or specialist software. For S+A, the value wasn’t simply faster rendering. It was the ability to test architectural decisions without losing confidence that the visualization still represented the original design.
“Every project is guided by clear objectives that we define with the client from the start. That means constantly testing ideas, optimizing materials and refining the design. Visoid allows us to explore those alternatives almost instantly. Instead of waiting days to visualize an option, we can discuss different solutions with the client almost immediately,” says Tavares.
An apartment block in Moita, where S+A sought to create variety in the modular design without undermining construction efficiency.
While the lifestyle headlines announce Cristiano Ronaldo and Georgina Rodrígues new property in Cascais valued at around 25 million euros — a figure that would make it Portugal’s most expensive private home — the reality is widely different for the midmarket housing in the country.
The median sale price of family housing rose 19.8% in the first quarter of 2026 compared to the same period in 2025, according to Portugal’s National Statistics Institute (INE), reaching €2,337 per square meter nationally in 2026 and €3,836 in Greater Lisbon. Across the EU, residential construction costs rose 5% in the first quarter of 2026 compared with the first quarter of 2025, according to Eurostat.
Unless your client is the five-time Ballon d’Or winner, that’s the backdrop many architecture firms are designing into right now. The budget gets tight, and somewhere between the design studio and the site, a material gets swapped, a detail gets simplified, and a concept gets diluted.
“One of the biggest challenges in residential architecture is finding the balance between variety and efficiency. We need, in as early a stage as possible, to predict and optimize construction systems without compromising the design and function of spaces. Visoid lets us do this while maintaining a high fidelity of the tested outputs. By testing different materials, configurations and solutions early on, we can make better decisions before the project moves forward,” says Tavares.
S+A now uses AI at two distinct moments in a project: during early design exploration, when architects evaluate different expressions, materials and atmospheres, and later in detailed design, when material substitutions, façade revisions and construction optimizations enter the conversation.
Every change can have both financial and architectural consequences. Traditionally, these discussions relied on technical drawings or lengthy rendering cycles, making it difficult for non-design stakeholders to understand the architectural consequences of seemingly small decisions.
S+A now uses AI as a design conversation tool, allowing its architects to compare material alternatives in real time, demonstrate how a different façade system changes the perception of a building or explain why a specific finish may ultimately undermine the project’s architectural intent.
In Sintra, S+A explore how to create a shared architectural identity amongst three distinct apartment towers.
That shift becomes most visible in three large residential developments currently underway around Lisbon, where visualizations were used to evaluate architectural decisions while there was still time to change them.
At Moita, the question was how to create variety without undermining construction efficiency. Located on the south bank of the Tagus River, the Moita project includes 240 apartments across 252,952 square feet (23,500 square meters) above ground.
Here, the challenge was developing a modular architectural language that could remain economical without becoming repetitive. By testing façade materials, finishes and recurring architectural elements from early massing models, the design team could explore how much variation was possible within a rational construction system.
At Sintra, the challenge was different: how to give three towers a shared identity without making them visually identical.
The project comprises three residential towers and 498 apartments across 554,341 square feet (51,500 square meters) above ground. Rather than committing to a single direction early, the architects evaluated numerous combinations of materials, lighting conditions and color palettes for the three buildings.
The process allowed the team to explore how relatively small changes could give each tower a distinct character while maintaining a coherent architectural language across the development.
At Bela Vista, AI became a way to test materiality and atmosphere earlier in the design process.
Closer to central Lisbon, the Bela Vista project has 211 apartments across 220,660 square feet (20,500 square meters) above ground. Here, the team used AI to study both exterior materiality and interior atmosphere well before detailed design begins.
Daytime and nighttime views, lobby interiors and façade treatments were evaluated as part of an iterative design process. Rather than treating visualization as the final presentation of a design, the architects could use it to ask questions: How does a material change the character of the building? What happens to an interior atmosphere under different lighting conditions? Does a façade treatment reinforce the architectural intent or work against it?
The result was a process in which architects and clients could align around design decisions earlier, while there was still time to adjust them.
For their Bela Vista design, S+A used the AI-generated renders as part of their explorative, iterative process before landing on any absolute choices for the building’s materiality.
When budgets get tighter, architectural quality is often the first thing to become negotiable. S+A believes AI changes that equation by making those trade-offs visible earlier, before they become permanent compromises.
According to S+A’s Marketing Manager Manuel Banazol, a set of twelve renderings that once could take up to a week can now be produced in a fraction of that time, with internal resource savings reportedly reaching 80–90% depending on the type of visualization.
But for the architects at S+A, speed is a consequence rather than the primary benefit.
“Avoiding AI today would be a mistake,” says Tavares. “These tools allow us to test more ideas, involve clients earlier in the process and make better-informed decisions together. Ultimately, they give us more time to focus on what really matters: the quality of the architecture.”
Over the past two years, AI has moved from an experiment to an integral part of the firm’s design process.
“We started trusting the technology when we realized it wasn’t just saving time — it was improving conversations with clients and helping us make decisions earlier,” says Tavares. “That’s when we understood it wasn’t simply a visualization tool anymore; it had become part of the creative process.
Architects: Want to have your project featured? Showcase your work by uploading projects to Architizer and sign up for our inspirational newsletters.
The post The Cost of Every Compromise: How Saraiva + Associados Brought AI Into Residential Practice appeared first on Journal.
Portugal needs more housing, but delivering it without compromising architectural quality is becoming increasingly difficult. Saraiva + Associados (S+A), a Lisbon-based architecture practice with 30 years of experience, has spent the past two years trying to close that gap using AI.
With more than 100 people on staff and operating across nine countries, S+A is the largest architecture-only practice of its kind in Portugal. Like many practices around the world, S+A began experimenting with AI out of curiosity. Now, the firm is integrating AI into the architectural decision-making process and helping clients understand the consequences of design choices before they become expensive compromises.
At first, the office ran a wider evaluation across multiple AI visualization tools and kept hitting the same issue: geometry drift. “The first tests we ran were crazy,” says Luís Barros, Chief Operating Officer and Architect at S+A. “A designer had a concept for a house, and by the end it had become something completely wild and different.” The tools were good at generating images, but they were bad at staying inside the idea.
“For us, it’s never about sacrificing design to reduce costs. It’s about designing within a realistic framework that balances architectural quality, budget and delivery time,” says Tiago Tavares, Senior Partner and Architect at S+A. If an AI tool changes the architectural concept simply to produce a visually attractive image, the rendering is not useful anymore.
Keeping the Architectural Concept Intact
Saraiva + Associados (S+A) architects Manuel Banazol, Luis Barros and Tiago Tavares
The S+A team didn’t give up and decided to set up an internal investigation group with twelve people testing AI tools against the firm’s actual workflow. One tester brought Visoid to the table, a tool that, unlike the others, held the line on geometry, followed the concept and rendered it. “We’re so careful about following our clients’ requests closely, it was easy to see that a tool like this made perfect sense for us,” says Barros.
Visoid is a rendering tool built specifically for architects. Using AI, it turns design models into high-quality visuals in minutes without the usual rendering workflows or specialist software. For S+A, the value wasn’t simply faster rendering. It was the ability to test architectural decisions without losing confidence that the visualization still represented the original design.
“Every project is guided by clear objectives that we define with the client from the start. That means constantly testing ideas, optimizing materials and refining the design. Visoid allows us to explore those alternatives almost instantly. Instead of waiting days to visualize an option, we can discuss different solutions with the client almost immediately,” says Tavares.
Designing When Every Square Meter Counts
An apartment block in Moita, where S+A sought to create variety in the modular design without undermining construction efficiency.
While the lifestyle headlines announce Cristiano Ronaldo and Georgina Rodrígues new property in Cascais valued at around 25 million euros — a figure that would make it Portugal’s most expensive private home — the reality is widely different for the midmarket housing in the country.
The median sale price of family housing rose 19.8% in the first quarter of 2026 compared to the same period in 2025, according to Portugal’s National Statistics Institute (INE), reaching €2,337 per square meter nationally in 2026 and €3,836 in Greater Lisbon. Across the EU, residential construction costs rose 5% in the first quarter of 2026 compared with the first quarter of 2025, according to Eurostat.
Unless your client is the five-time Ballon d’Or winner, that’s the backdrop many architecture firms are designing into right now. The budget gets tight, and somewhere between the design studio and the site, a material gets swapped, a detail gets simplified, and a concept gets diluted.
“One of the biggest challenges in residential architecture is finding the balance between variety and efficiency. We need, in as early a stage as possible, to predict and optimize construction systems without compromising the design and function of spaces. Visoid lets us do this while maintaining a high fidelity of the tested outputs. By testing different materials, configurations and solutions early on, we can make better decisions before the project moves forward,” says Tavares.
S+A now uses AI at two distinct moments in a project: during early design exploration, when architects evaluate different expressions, materials and atmospheres, and later in detailed design, when material substitutions, façade revisions and construction optimizations enter the conversation.
Every change can have both financial and architectural consequences. Traditionally, these discussions relied on technical drawings or lengthy rendering cycles, making it difficult for non-design stakeholders to understand the architectural consequences of seemingly small decisions.
S+A now uses AI as a design conversation tool, allowing its architects to compare material alternatives in real time, demonstrate how a different façade system changes the perception of a building or explain why a specific finish may ultimately undermine the project’s architectural intent.
Three Examples of Cost Control as Part of the Design Process Itself
In Sintra, S+A explore how to create a shared architectural identity amongst three distinct apartment towers.
That shift becomes most visible in three large residential developments currently underway around Lisbon, where visualizations were used to evaluate architectural decisions while there was still time to change them.
1. Moita: Making Repetition Less Repetitive
At Moita, the question was how to create variety without undermining construction efficiency. Located on the south bank of the Tagus River, the Moita project includes 240 apartments across 252,952 square feet (23,500 square meters) above ground.
Here, the challenge was developing a modular architectural language that could remain economical without becoming repetitive. By testing façade materials, finishes and recurring architectural elements from early massing models, the design team could explore how much variation was possible within a rational construction system.
2. Sintra: Giving Three Towers Different Identities
At Sintra, the challenge was different: how to give three towers a shared identity without making them visually identical.
The project comprises three residential towers and 498 apartments across 554,341 square feet (51,500 square meters) above ground. Rather than committing to a single direction early, the architects evaluated numerous combinations of materials, lighting conditions and color palettes for the three buildings.
The process allowed the team to explore how relatively small changes could give each tower a distinct character while maintaining a coherent architectural language across the development.
3. Bela Vista: Testing Atmosphere Before Committing to It
At Bela Vista, AI became a way to test materiality and atmosphere earlier in the design process.
Closer to central Lisbon, the Bela Vista project has 211 apartments across 220,660 square feet (20,500 square meters) above ground. Here, the team used AI to study both exterior materiality and interior atmosphere well before detailed design begins.
Daytime and nighttime views, lobby interiors and façade treatments were evaluated as part of an iterative design process. Rather than treating visualization as the final presentation of a design, the architects could use it to ask questions: How does a material change the character of the building? What happens to an interior atmosphere under different lighting conditions? Does a façade treatment reinforce the architectural intent or work against it?
The result was a process in which architects and clients could align around design decisions earlier, while there was still time to adjust them.
The Productivity Gains Are Significant
For their Bela Vista design, S+A used the AI-generated renders as part of their explorative, iterative process before landing on any absolute choices for the building’s materiality.
When budgets get tighter, architectural quality is often the first thing to become negotiable. S+A believes AI changes that equation by making those trade-offs visible earlier, before they become permanent compromises.
According to S+A’s Marketing Manager Manuel Banazol, a set of twelve renderings that once could take up to a week can now be produced in a fraction of that time, with internal resource savings reportedly reaching 80–90% depending on the type of visualization.
But for the architects at S+A, speed is a consequence rather than the primary benefit.
“Avoiding AI today would be a mistake,” says Tavares. “These tools allow us to test more ideas, involve clients earlier in the process and make better-informed decisions together. Ultimately, they give us more time to focus on what really matters: the quality of the architecture.”
Over the past two years, AI has moved from an experiment to an integral part of the firm’s design process.
“We started trusting the technology when we realized it wasn’t just saving time — it was improving conversations with clients and helping us make decisions earlier,” says Tavares. “That’s when we understood it wasn’t simply a visualization tool anymore; it had become part of the creative process.
Architects: Want to have your project featured? Showcase your work by uploading projects to Architizer and sign up for our inspirational newsletters.
The post The Cost of Every Compromise: How Saraiva + Associados Brought AI Into Residential Practice appeared first on Journal.