Anton Juric: AI Won't Replace the Engineer — It Will Change Where Their Judgment Matters Most

Most people would probably assume that the systems inside a modern building are already designed by sophisticated software. Anton Juric says that assumption confuses digital drafting with automated engineering. As President and Co-Founder of Endra, he has spent the past two years helping major engineering consultancies evaluate whether AI can move from demonstration into actual project work. His perspective comes from the commercial side of that transition, where adoption depends on how well a system fits the realities of engineering practice.

Juric argues that software has changed the way engineers draw without fundamentally changing how much of the underlying design work still requires manual effort. Tools such as Revit moved the drafting board onto a screen, but many calculations and engineering decisions remain dependent on people working across disconnected systems. He sees generative design as a more significant change because the software can begin producing portions of the design itself. That shifts the discussion from how engineers use digital tools to how their role changes when software can take on more of the production work.

"For years, software helped the engineer document the work more efficiently," Juric said. "What is changing now is that the system can participate in producing the design itself. That changes where the engineer spends time and where professional judgment becomes most valuable."

The distinction is especially important in a field governed by physical constraints. Building systems have to comply with codes and technical standards, and errors can have serious consequences. Juric does not believe general-purpose AI can simply be introduced into engineering workflows without being adapted to the discipline. His view is that AI for engineering has to be developed around domain-specific requirements and alongside the professionals who understand how those requirements apply in practice.

"Engineering software has to understand the context in which the work is being done," he said. "A generic system may be useful for many things, but engineering decisions are tied to rules, calculations, and physical realities. The technology has to be built around those conditions."

Juric's role at Endra has placed him directly inside the conversations that determine whether firms are willing to adopt that technology. He leads the company's commercial operations, including sales, partnerships, and international expansion, and says he has personally negotiated agreements with some of the world's largest engineering consultancies. Those discussions have given him a close view of what enterprise buyers actually examine before placing AI into production.

One lesson has been that engineering firms are less interested in the label attached to the technology than in what it changes operationally. Juric says they want to know whether a system can save time, fit existing workflows, and reduce the burden of repetitive work without compromising the quality expected from professional engineers. That has shaped his broader view of how technical founders should approach enterprise customers. His advice is to build deeply for a specific industry, stay close to the people using the product, and focus the commercial conversation on results rather than novelty.

"Engineering firms are not buying AI because the term itself is persuasive," Juric said. "They want to know what changes in the delivery of a project. The commercial conversation becomes much more serious when you can talk about the work they need to complete rather than the technology you want them to adopt."

Endra's adoption figures give Juric a concrete basis for that view. The company has signed around 30 enterprise clients, and Juric says every client validation conducted by Endra has resulted in a signed contract. For this story, the significance of that figure is less about company growth than what happens after engineering firms test the technology. Juric sees the conversion from validation to contract as evidence that some firms are finding enough practical value to continue into commercial use.

That movement from testing to production is part of a broader change Juric believes is underway in architecture, engineering, and construction. For decades, digital tools primarily helped professionals execute work they had already decided how to do. Generative systems introduce a different possibility by producing design outputs that an engineer can review, refine, and ultimately approve. The distinction places greater importance on the boundary between automated production and professional responsibility.

Juric believes that the boundary should remain clear. His argument is that automation is most useful when it expands the amount of meaningful engineering work a qualified professional can complete. Repetitive documentation and design tasks can consume substantial time, while the engineer's expertise is most valuable in decisions that require technical judgment and accountability. In that model, AI changes the allocation of work without removing the engineer from the process.

"The value is in giving engineers more room to do the work that actually depends on their expertise," Juric said. "If software can handle more of the repetitive production, the engineer can spend more time reviewing, deciding, and solving the parts of the project that require judgment."

His experience as a second-time founder has also influenced how he approaches that transition commercially. Juric says he learned from his earlier entrepreneurial experience that technical quality alone does not guarantee adoption, particularly in enterprise markets. Distribution matters, and customers have to see a clear operational reason to change established ways of working. He also says he now makes decisions faster and is more willing to hire ahead of growth than he was earlier in his career.

Juric expects generative design to become far more common inside major engineering firms over the next several years. He believes AI-generated MEP design could become standard practice within three years, particularly as consultancies look for ways to increase delivery capacity and make better use of scarce technical talent. That remains his forecast, but it is informed by the adoption discussions he is already having with firms evaluating the technology today.

"The profession is not becoming less important because software can do more," Juric said. "The role becomes more focused on judgment, responsibility, and the decisions that require an engineer to understand the whole system. I think the firms that adapt to that shift early will change what their engineers are able to deliver."