Takeaways on the Future of Construction from Autodesk University
Early signals on how AI will reshape design, data, and construction workflows
Originally published by Bhragan P. on Last Week in ConTech on December 18, 2025
This article is written by Erin Khan. She is an established AEC technology leader who provides technology and innovation services to both contractors and construction technology startups at her consultancy, Erin Khan Consulting. Prior to founding EKC in 2023, Erin served as the National Director of Construction Solutions for Suffolk Construction.
Autodesk University has always been one of the most important events in AEC.
When I was the National Director of Construction Solutions at Suffolk Construction, attending conferences was valuable to understand where the market was heading and how emerging technologies might reshape project delivery.
Even though this conference is centered on Autodesk and its product suite, their roadmap offers one of the clearest signals of how the industry's digital ecosystem may evolve. Hearing how a major software provider interprets market trends helps place our own technology decisions in context. It shows what is feasible today, what is emerging, and when it might be realistic to pilot new solutions.
Now that I work as an independent ConTech consultant, I attend conferences for the same reason.
The insights are valuable because they highlight not only what Autodesk is building, but also how the broader sector is shifting. Here are my key takeaways from this year's event and what they signal about the future of construction.
Contents
-
Takeaway 1: AI Is Here, but It Still Needs Work
-
Takeaway 2: Data Security is Everyone's Problem
-
Takeaway 3: We Need More Content Creators for Education
-
Takeaway 4: The Trades are a Massive Untapped Opportunity
-
Final Takeaway: Ask the Right Questions Before Adopting a New Solution
Takeaway 1: AI Is Here, but It Still Needs Work
One of the biggest announcements at the conference was Autodesk's introduction of 'neural CAD'.
It's a new category of generative AI foundation models trained to understand and reason about CAD objects.
They outlined two models being:
-
Neural CAD Geometry
This is a generative model that creates design from text prompts and spatial constraints. Users can then adjust and refine the generated geometry. -
Neural CAD for Buildings
This is a model enabling architects to quickly transition between early design concepts and more detailed building layouts and systems, translating between a conceptual architectural massing model and building floorplan layout.
What was impressive was that the keynote speakers completed live demos of these models. Having done my fair share, the scope for failure or technical difficulties is large, proving the validity of this development.
The truth, however, is that while there is potential for AI and the examples were impressive, it just didn't seem like it was quite there yet. I have a healthy skepticism that intense, real-world applications would not be as smooth or seamless of an experience for the user, and it brought to mind how the early internet was. Initially it was clunky (anyone remember AOL dial up?) but over time and with iterations, it became a new standard of how we operate and exist in the world.
For me as an attendee, I'd like to understand the adoption and see the models stress tested against critical use cases. Yes - it looked cool, but can it live up to the hype? My gut says yes, but it needs more time to mature.
What was more interesting was what the announcement says about the future of design.
Autodesk sees design shifting toward a prompt driven workflow, where initial concepts come from natural language inputs and designers spend more time editing, refining, and evaluating AI generated options. This places more capability directly in the hands of engineers and designers who historically relied on drafters or modellers for early iterations.
Additionally, Autodesk envisions being the 'all in one' tool; becoming a dynamic and intelligent platform to meet a wider range of project needs.
Takeaway 2: Data Security is Everyone's Problem
One of the most eye opening sessions I attended was called 'Trust Enthusiasts Unite.'
It was a community meetup of cybersecurity and data experts. I was the only person from operations in the session, and what was evident was how construction professionals don't really think about data security. The (risky) assumption is that IT will handle it.
But as AI tools become embedded in day to day work, the risk profile is shifting - significantly. End users now have far more influence over where data goes, how it is handled, and the resulting risk level created.
For example, an engineer may upload a confidential report to ChatGPT to speed up their workflow without recognizing the security implications. They rarely consider the security implications, the data retention policies, or whether the platform is approved for that type of information. With AI, the guide rails are thinner and the likelihood of accidental exposure increases.
The question is no longer just, 'How does IT vet this solution prior to adoption?' It's also, "How do we reduce risk through behaviour and standards?"
Right now, the industry has no common approach.
Companies are creating their own AI policies, sometimes limited to a paragraph in an employee handbook that says "don't upload sensitive data." It's just not enough.
My take is that AEC needs a structured framework for safe AI usage, similar to how BIM standards define Levels of Development or how OSHA sets rules for onsite safety. Without shared definitions, training, and expectations, every company is reinventing governance in isolation.
Want to chat with me on this topic? Please send me an email at: erin@erinkhanconsulting.com or send me a message via LinkedIn.
Takeaway 3: We Need More Content Creators for Education
Most training content for AEC professionals kind of sucks.
It's mostly dry videos that aren't that exciting, or are really long and not the best quality. As we discussed in the Content Creators Breakout session, most people are finding content for self learning either on:
-
LinkedIn Learning
-
YouTube
-
Company Learning Management Systems
There's no single source of high quality, practical learning content, and nobody in the breakout room could name one platform they consistently rely on.
The result is a knowledge bottleneck.
Most expertise lives in people's heads or inside proprietary company systems. When information is shared, it usually stays internal. Little of it reaches the broader industry or the next generation of talent.
This lack of accessible content creates two problems.
1. It slows down adoption of new tools and best practices.
If the industry wants to scale AI, robotics, or new workflows, people need to understand them. Right now, the learning curve is too steep and the materials too outdated.
2. It hurts our ability to attract early career professionals.
People learn on platforms like YouTube, TikTok, and Instagram. That is where attention is, and if we want the industry to be compelling, we need content that is clear, engaging, and easy to find.
Other industries have solved this by empowering practitioners to share knowledge publicly, and more recently, expanding to influencer partnerships.
This is where individuals who have captive, younger audiences share educational content, helping to connect industry roles to modern cultural trends. Examples include:
-
The rise #BlueCollar Trades
Creators like Czumak-Abreu, share day to day electrician work and earn over $200,000 annually through brand partnerships with companies such as Klein Tools and Carhartt. -
John Deere's #FarmTok
Recognizing growing curiosity about agriculture alongside a lack of understanding, John Deere created content aimed at Gen Z audiences who are increasingly disconnected from farming, highlighting modern technology, innovation, and new career pathways within agriculture. -
The creation of US Army Esports
After failing to meet recruiting targets, the US Army sponsored an esports team to reach younger gamers. The program has since been widely cited as one of their most effective modern recruitment tools.
Construction can do the same, but only if more people step forward to teach, publicly document their expertise, and be open to creative collaboration opportunities.
Takeaway 4: The Trades are a Massive Untapped Opportunity
One of the clearest gaps highlighted at Autodesk University is how overlooked the trades remain in the current wave of construction technology.
Most startups are building for General Contractors, and the Subcontractors are told to just follow the GC. There's an entire layer of trade specific workflows, which, if these were optimized and integrated, we would be able to increase efficiency, reduce rework, and catch problems earlier.
This became obvious during Augmenta's session. Their product, which automates electrical system design, removes the repetitive, detail heavy tasks that bog down modellers. Instead of drawing lines or placing components one by one, users can be more creative, focusing on higher level design decisions and system optimization.
It's made possible by pairing AI with humans, and is the practical application of the vision Autodesk described with neural CAD, placing more modelling capability in the hands of engineers.
To unlock this opportunity at scale, the industry needs more trade experts partnering with startups or even founding their own. The next wave of ConTech breakthroughs won't come only from GC workflows or corporate model-based coordination. They will come from solving the detailed, high frequency tasks that shape every project but rarely get the software investment they deserve.
Final Takeaway: Ask the RIght Questions Before Adopting a New Solution
When I go to these conferences, I feel optimistic and excited about the future of our industry. Everything is presented as a glossy vision for the future, changing how we build.
But it's also easy to get swept up in the excitement and lose sight of what matters most: choosing tools that work for your teams in the real world.
We can do so by understanding what the right questions are to ask when evaluating the new startups which are emerging.
And a mental shift I emphasize more often is this: to not think about a startup as a pure technology purchase, but rather as a partner in workflow enablement and problem solving.
It's because startups act as the shared R&D of the industry.
When you buy a product, you're adopting a workflow that has been shaped by hundreds of conversations with end users like field teams, designers and project managers. In many cases it gives the mid market firms access to the operational systems that are often limited to the enterprise layer.
That's why due diligence needs to go beyond features or demos. You're assessing whether the underlying workflow matches how your teams operate (or should operate), where your bottlenecks are, and where you want to build capability.
The right question isn't, "Can the product do this?"
It's, "Does this workflow make us better?"
Given that, make sure you ask the right question to ensure the product and the workflow it represents align with your specific needs.
Autodesk University is always a glimpse into the future.
This year made one thing clear: AI's evolution is only just beginning; the real shift ahead will be magnitudes greater than what we see today. The companies that will lead the next era of construction are those that lean into this moment - selecting solutions that truly matter, reshaping their workflows, and empowering their teams to use AI with skill and responsibility.
Erin Khan is an established AEC technology leader, with over a decade of combined experience in construction operations, data and process analysis, and software implementation. Founding Erin Khan Consulting in 2023, she currently provides technology and innovation services to both contractors and construction technology startups.
Prior to founding EKC, Erin served as the National Director of Construction Solutions for Suffolk Construction, overseeing national technology and innovation operations. Erin holds a B.S. in Civil Engineering from USC, a certificate in Business Analytics from The Wharton School, and is EIT, LEED AP BD+C, and OSHA-10 certified.