From Automation Leader to Solution Architect at Autodesk with Adam Sheather

The Digital Construction Podcast recently hosted Adam Sheather, Solution Architect at Autodesk, for a deep dive into automation, computational workflows, and the future of AI in construction. Drawing on an unconventional career Adam offered practical perspectives on how AEC firms can move past simply developing "cool" scripts to genuinely align operational initiatives with business funding and objectives. He shared insights on the true definition of automation, common pitfalls in workflow development, how value differs for designers versus contractors, and where the next big opportunities lie in the "AI adjacent space."

The Role of a Solution Architect at Autodesk

The higher purpose role of a solution architect is truly about bringing business, operations, and IT strategy together. Where disconnects often occur is when operations teams develop automation scripts or solutions, but fail to align them with the business side, which focuses on finance, targets, and KPIs. The role spans a vast spectrum, from planning software development projects that solve customer integration problems to setting higher-level road maps that ensure initiatives are aligned with core business goals.

Demystifying Automation: Reducing Low-Value Tasks

Adam emphasized the need to move away from the idea that it must involve complex code or "magic". At its core, automation is simply reducing low-value tasks through a computerized process, allowing users to spend less time on repetitive work. A simple example shared was using advanced Excel functions to automatically calculate a bill of materials for scaffolding, reducing the time spent on tendering from three hours to just 30 minutes per project, without needing any complex code. This focus on reducing time on low-value tasks applies whether the solution involves simple scripting or a complex system.

The Missed Opportunity: Automating the "Boring" Work

A common mistake in the industry is focusing efforts on automating things that are "really cool or interesting," rather than tackling the boring, repetitive tasks that offer the highest efficiency gains. Adam described projects where immense time savings could be achieved through simple file management tools, but these tasks are often overlooked because people don't see the value. He advises teams to focus on solving those boring tasks first.

Building Computational Workflows: Know Your Tools

When approaching the development of computational workflows, which are essentially automations, whether sitting in Rhino, Grasshopper, or Dynamo, the priority must be understanding the tool's capabilities and the true value of the problem being solved. A frequent pitfall is seeing users write complex routines in tools like Dynamo to automate something that Autodesk Revit can already accomplish natively, or adding third-party libraries to perform calculations for booleans when a geometry kernel already exists in-house. Adam finds that the complexity in end-to-end workflows often stems from inconsistent business processes and the "people piece," rather than technical limitations.

Tailoring Automation Value: Designers vs. Contractors

The value derived from automation fundamentally differs based on the business model. For a builder working on a fixed-price contract, any workload that saves money automatically translates to profit. Adam recounted a project where redrawing a complex formwork design in Revit highlighted that 85% of the proposed components were custom; redesigning it to reduce custom components down to 15% saved the project $4 million in just two weeks of work, a clear, tangible outcome.

In contrast, design firms operating on a time and materials basis often find the value more elusive. While saving 25-30% on drafting time is achievable, that time often gets reallocated rather than resulting in a lower charge to the client. If this saved time is not intentionally recaptured and reallocated into high-value activities for the client (like enhanced engineering or simulations), the project itself might not see measurable value, leading to frustration and cancelled automation initiatives. Furthermore, introducing new processes (like attaching complex data attributes for the client/asset owner) can face resistance from users if the automation does not offer them a meaningful benefit or help their immediate design task.

The Next Frontier: The AI Adjacent Space

Looking ahead, Adam believes the biggest opportunities for automation lie in the AI adjacent space. This involves leveraging existing, powerful tools like Large Language Models (LLMs), such as ChatGPT or Google Gemini, to take actions on a user's behalf or to recall information. Rather than firms building custom ML models from scratch, the immediate value is in utilizing these models for practical, construction-specific tasks. This includes using vision systems to read drawings, interpret data, or perform automated QA checks, such as verifying if every drawing has a North arrow. This use of AI for reading documents, drawings, and images to translate a meaningful outcome can have significant value.

Key Takeaways

• Automation is fundamentally about reducing low-value tasks using a computerized process, regardless of whether that is fancy Excel or complex code.
• The essential first step in any major automation journey is ensuring business, operations, and IT strategy are in sync to secure necessary funding and resources.
• Avoid the trap of automating "cool" features; prioritize solving the repetitive, boring tasks (like file management or basic data entry) as they often yield the largest efficiency gains.
• When developing computational workflows (e.g., in Revit or Dynamo), ensure there is an in-depth understanding of the underlying software's capabilities to avoid unnecessarily rewriting functions.
• The primary challenge in complex, end-to-end computational workflows is inconsistent business processes and the human element, not the technology itself.
• For designers and engineers, time saved by automation must be intentionally recaptured and allocated to higher-value activities for the client, otherwise the benefit is lost to the project.
• The next big opportunity lies in the AI adjacent space, leveraging existing LLMs for automated QA checks, data extraction from drawings, and digitizing information.