Building the foundations for AI

For most architecture firms, the AI conversation starts with a question: what can it do for us? At SAMOO Architects & Engineers, that question arrived after years of preparation.

One of Korea’s largest architecture and engineering firms, with over 1,000 professionals and roughly 8,000 projects to its name, SAMOO has spent years building the digital foundations that make AI useful in practice, not just promising in theory.

Yoon Jong-geun, who leads innovation initiatives within SAMOO's AI Transformation (AX) Team, describes that three-stage journey as building systems and data foundations first, then improving work through automation, and now exploring how AI can change how people collaborate, decide, and solve problems.

SAMOO's AX team has worked closely with business units across the company to build the tools, systems, and standards behind more connected ways of working. Their first task was foundational: a BIM-based design framework, shared standards and libraries, and consistent conventions for how drawings, sheets, and project data were structured firm-wide.

With that infrastructure in place, the team turned to the next challenge: repetitive work consuming designers' time at scale. SAMOO developed Revit add-ins such as SBE (SAMOO BIM Extension) and AutoDrawings, identifying field challenges and defining features internally, and building solutions with external specialist partners. Some of these tools removed internal bottlenecks; others grew into standalone products, including ARCHITEC'TWIN, a digital twin platform for post-occupancy building management, which won the Best of Best at the 2024 Red Dot Awards.

When a large project team approached SAMOO's AX team, it looked at first like a straightforward automation task.

Exporting construction drawings from Revit on large-scale projects had become a slow, friction-laden process that teams simply absorbed. With drawing sets running into the hundreds or thousands of sheets, local machines struggled under the load. So teams divided sheets across multiple computers and ran exports in overlapping batches. When a machine froze mid-run, hours of work had to be repeated from scratch.

"There were serious failures mid-export," Yoon recalls. "We had to find a way to solve this. We thought we were starting an automation project."

The fix depended on something SAMOO already had: standardized sheet templates and project structures from its earlier BIM work. With every project now having the same underlying logic, the team could design a single automated PDF export process that would reliably hold across projects, rather than a fragile script tailored to one.

Working with Autodesk IS Consulting, the team built a cloud-based solution using Autodesk Platform Services (APS), Autodesk's toolkit for automating and extending design workflows in the cloud, together with Autodesk Forma (formerly Autodesk Construction Cloud). The solution ran the Revit engine remotely and removed the hardware constraint entirely. A full-day export process came down to 2.3 hours.

Beyond the workflow improvement, a more significant shift was happening inside the team. AI helped them navigate the learning curve of APS, explore technical possibilities faster, and close the gap between planning intent and what could actually be built.

Returning to the export workflow, the team asked a new question: if local hardware was no longer the constraint, what else was possible? The answer was parallel processing. Instead of running jobs sequentially, they distributed workloads across 48 parallel cloud jobs. The result: 624 sheets exported in 30 minutes, a reduction of more than 94% from the original full-day process.

The most lasting change wasn't in the workflow. It was in how the team worked.

Before this project, SAMOO's AX team operated with clear professional divisions: planners planned, developers built, project managers coordinated. As AI became a shared capability across the team, rather than a tool owned by one function, those boundaries began to blur.

"Until last year, I was responsible for planning and project management," says Yoon. "Now I plan, develop solutions, and manage projects end to end. One person can now take on roles that previously would have been divided across a team."

Yoon sums up this shift in a phrase: "One person, powered like a team.". It is what becomes possible when one person can move fluidly across the full cycle of problem-solving, without waiting on handoffs that slow things down.

The firm has built governance to match. An AI Steering Committee meets weekly at C-level to monitor use cases and coordinate policy. Teams surface problems through an internal AI use case system, and the most promising ideas become shared solutions.

SAMOO’s PDF export workflow produced a measurable result: eight hours down to 2.3, then to 30 minutes. But the more durable outcome is what that result revealed. A drawing export bottleneck became a cloud-based workflow, then a new way of building solutions with AI.

Yoon says the AX Team is now applying the same thinking to its next challenge: making project knowledge easier to reuse. As more design information moves into the cloud, the team is exploring knowledge graphs and ontology-based structures that can help teams build on what SAMOO already knows and use AI to solve new problems faster.

Yoon sees the APS project as the start of broader collaboration with Autodesk across cloud and AI technologies, consulting support, and joint AX workshops.