For as long as I’ve been thinking about learning and development in AEC, the guiding ambition has been clear — deliver the right knowledge to the right person at the right time. It sounds simple, but anyone who’s managed L&D in a design firm knows how complicated that really is.
When people have timely access to the right resources, they are better prepared for the projects in front of them, more confident in their roles, and more likely to grow into future leaders. But timing matters. If knowledge arrives too early—before there’s a chance to apply it on a real project—it often doesn’t stick. If it comes too late, you can miss a golden learning opportunity.
That’s why the topic of automation drew so much energy in our recent conversations with private beta cohorts about Synthesis LMS. Could technology finally solve the “right person, right time” problem? What if assignments, reminders, and learning paths could run on autopilot?
It’s an exciting vision. But as we talked it through, a more nuanced picture emerged. Automation will help—sometimes enormously. However, you can’t automate all delivery of learning content. Just as we’ve learned with AI, automation without a human in the loop can quickly go off in wild directions. The real challenge isn’t simply automating more, but deciding when and how to keep humans close to the process—ensuring assignments come with judgment, empathy, and context.
The Air Traffic Control Problem
One of the metaphors that has guided us in designing Synthesis LMS is air traffic control. In every firm, there are many learners, many learning resources, and countless situations where knowledge could make a difference. The real art is in connecting the right learning opportunity to the right person at just the right moment—or even slightly ahead of when they need it, which can be the most powerful timing of all.
That responsibility often falls to L&D teams. In some firms, it lives in HR; in others, with departmental managers or individual project leaders. Sometimes all of the above. Whoever plays the role, they are in effect the air traffic controllers—tracking people, courses, learning paths, and the situations that call for them. Much of this orchestration happens informally, held in people’s heads or cobbled together in spreadsheets, with everyone doing their best to keep things moving.
Automation offers a chance to relieve some of that burden. Instead of manually tracking, managers could set rules and let the system route learning automatically. Promotions, new hires, compliance deadlines—these are clear, predictable triggers where automation can help ensure that learning content is delivered reliably and on time.
As we explored the possibilities of automation with our private beta cohort, it became clear that not every “flight plan” can—or should—be created by software. Some assignments will always require the judgment, empathy, and context that only humans can provide.
In other words, automation isn’t just a technology question—it’s a design problem. How do we design systems that reliably connect the right knowledge at the right time to the right person, without overwhelming them, or losing the human touch?
Low-Hanging Fruit: Smarter Assignments Through Employee Profile Fields
In version one of Synthesis LMS, we’ve started with practical, low-hanging fruit — automation that builds on information firms already keep in employee profiles. Titles, departments, locations, years at the firm, and custom fields can all become the basis for smarter assignments.
There are three types of assignments in Synthesis LMS:
Basic assignments: Directly assigning a course or learning path to an individual or group. Simple, manual, and useful in day-to-day situations.
Smart assignments: Rules-based assignments using employee profile fields. These can be one-time (“assign to everyone currently in the Structural Engineering department”) or ongoing (“any time someone is promoted to Project Manager, assign the Project Management Essentials path”).
Global assignments: Firmwide distribution, well-suited for compliance training or company-wide initiatives.
Even this first layer of automation goes a long way toward the goal of connecting the right knowledge with the right person at the right time. New hires can automatically move through onboarding. Newly promoted managers can receive leadership training right away. Compliance requirements can be met reliably without manual tracking.
Of course, not every learning need can be met this way. But for many firms, these simple tools will feel like real relief. They will take pressure off the teams who have been managing learning assignments manually, and make structured development feel less daunting for firms that haven’t had a system at all.
More Low-Hanging Fruit: Automating Staggered Assignments in Learning Paths
A Learning Path is a structured sequence of courses designed to guide employees through a series of topics or competencies. Learners can be assigned to a path, receive a recommendation, or discover and self-enroll directly.
In version one of Synthesis LMS, learning paths can also be automated so that courses are delivered gradually — over days, weeks, or months — rather than all at once. This helps avoid overloading new hires on day one. New employees need space to absorb onboarding information at a human pace. A project manager in training is better served by a steady sequence of learning rather than being overwhelmed all at once. Even in technical areas like Revit 101 or Introduction to Project Management, staggering content gives people time to practice, apply, and retain what they’ve learned before moving on.
This approach balances the efficiency of automation while honoring the human limits of attention and absorption. It also respects the reality of billable work, ensuring that learning is woven into people’s schedules instead of competing with them. For managers and L&D leaders, staggered assignments remove the burden of remembering when to release each course—the system takes care of pacing, while managers and mentors stay focused on providing context, encouragement, and support.
Can We Take Automation Further? The Jury’s Still Out
Beyond the basics, there are scenarios where automation could, in theory, go much further. These ideas are exciting to imagine, but it’s not yet clear whether they’re the right path—or whether the trade-offs will make sense in practice. For example:
A failed phishing test could automatically trigger a security refresher.
Products like Guardian for Revit or Bimbeats, which track inefficiencies and trends in the way employees model, could automatically suggest tutorials at just the right time
Deltek Vantagepoint could recognize that someone has been assigned to their first behavioral health project and surface a learning path tailored to that market.
An assessment or quiz could automatically assign follow-up courses, creating a more personalized learning path.
At first glance, these possibilities sound powerful. They could save managers time, provide just-in-time guidance, and reduce errors. They could also make learning feel more relevant, because it would be tied directly to the work at hand.
At the same time, our private beta cohort reminded us that this space is full of uncertainty. The data we’d need often doesn’t exist, isn’t reliable, or doesn’t capture important nuances. And even when it does, automating L&D without a human in the loop can miss the point. A system may know what happened, but it takes a person to explain why it matters and how it fits into the bigger picture.
What Can’t Be Automated: The Human Layer That Will Always Matter
There are dimensions of learning that no system will ever replace.
A mentor encouraging a young architect to explore mass timber, even before they have a project in that space.
A project manager noticing it’s someone’s first time detailing a roof or the first time doing a site visit.
A leader suggesting project management courses to a designer considering a career pivot.
Sometimes, the difference is not the content itself but the context in which it arrives. One example raised in our beta cohort was an employee being assigned to their first Kaiser Permanente project. In theory, a resource planning system could detect the project staffing decision and automatically push a “Welcome to Kaiser” course. But what if the person hasn’t yet been told they’re moving to that project? Without the human conversation first, the assigned course would feel disconnected, even disorienting. What makes the difference is the manager who says, “I know this is your first Kaiser project. Here’s some background that will help you feel prepared.” The course may live in the LMS, but the meaning comes from the relationship.
The same is true in technical training. Imagine working in Revit, thinking you’re doing fine, when suddenly a course on “detailing tips and tricks” is automatically assigned to you out of the blue. Without context, it could feel like a judgment. Or even a mistake. Far better would be a BIM manager—perhaps noticing signals from a tool like Guardian or Bimbeats—sitting down and saying, “Hey, it looks like you might be struggling with this part of the workflow. Is everything okay? I recommend this course—it’ll give you some basics that can make things easier.” The content is the same, but the empathy, timing, and communication make all the difference.
These kinds of moments happen in conversations, not systems. They depend on judgment, relationships, and intuition. They are how culture is transmitted. And they are too granular, too nuanced, or too aspirational to be captured in a data field.
This doesn’t mean technology has no role. A platform like Synthesis LMS can make it easy for mentors and managers to turn these conversations into concrete learning assignments or recommendations, helping ensure that good intentions lead to action. But it’s the human in the loop that gives learning its meaning, context, and sense of care.
Even in these areas that can’t truly be automated, there may be ways for technology to play a supportive role. Instead of auto-assigning a course out of the blue, a system like Synthesis LMS could nudge a manager, mentor, or HR lead with: “This is Denise’s first Kaiser project—consider assigning the Kaiser onboarding course,” or “It looks like Chad may be struggling with detailing in Revit—here’s a resource you might suggest.” The assignment itself still comes from a person, but automation can surface the right moment for the assignment and even use AI to propose language to help frame the conversation. It’s a future possibility worth exploring—one that uses automation not to replace judgment, but to strengthen it.
Assignments vs. Recommendations
One of the insights that came through strongly in our beta conversations was the importance of distinguishing between assignments and recommendations. In Synthesis LMS, we’ve intentionally built space for both—because they serve different purposes and carry different weight.
Assignments are binding. They come with authority, deadlines, and accountability. They are essential when the firm needs to ensure something gets done—whether it’s compliance training, onboarding for new hires, or role-specific development for project managers. Assignments can be made manually, but they can also be automated. Much of this issue has been about exploring those possibilities: when automation can help, when it should be tempered, and when a human in the loop is still needed to provide context.
Recommendations are different from assignments. They are always human-to-human, and they are not designed to be automated. Any employee can recommend a course or learning path to any other employee. These recommendations are non-binding—they don’t carry deadlines—but they create opportunities. A recommendation might be a mentor nudging a younger colleague toward a resource, a peer sharing something they’ve found helpful, or a manager encouraging curiosity in an area outside someone’s current role. Recommendations are at their best when they are responsive, personal, and even anticipatory of someone’s aspirations.
By supporting both structured assignments and open recommendations, Synthesis LMS helps firms balance the possibilities of automation with the limits of what can be automated. Assignments create efficiency and accountability where consistency matters. Recommendations preserve the relational, serendipitous layer of learning that will always depend on people. Together, they create a system where technology can amplify—but never replace—the human spark at the heart of learning.
The Limits of Automating AEC L&D Go Beyond Technology
The limits of automation are not just technical. Some are cultural, some are relational, and some are simply human.
Data limits: Not all the data we would need for perfect automation exists in today’s systems, and the data which do exist aren’t always reliable enough to drive learning decisions. Even if the right data did exist and were perfectly reliable, the cost of capturing, cleaning, and maintaining that data may not be worth the return. Sometimes the effort required to enable a given automation outweighs the benefit it delivers.
Context limits: Automation struggles to communicate why a course is being assigned. Is it developmental, a sign that the firm is investing in your growth? Is it corrective, a form of performance feedback? Or is it simply exploratory, a way of encouraging curiosity? Without a human explaining the intent, assignments can easily be misinterpreted—sometimes in ways that undermine the very learning they were meant to support.
Relationship limits: Learning often flows best through trust—between manager and employee, mentor and mentee, peer and peer. That relational context shapes how a course is received, and no system can replicate it.
Learning limits: Even when automation delivers the right course at the right time, real learning often requires practice and social connection. Role-playing a difficult conversation, shadowing someone on a job site, or working through a Revit scenario with an expert all benefit from the richness of human instruction or mentorship. In education circles this is sometimes called the “flipped classroom”: the basics can be automated and delivered on-demand, but deeper learning happens when people engage together—asking questions, hearing others’ perspectives, and learning alongside peers with a trusted subject matter expert in the room.
In the end, the boundary of automation isn’t just about what’s technically possible. It’s also about what’s wise. Just because we can doesn’t always mean we should. That’s why keeping a human in the loop is so important: it ensures that automation supports learning without stripping away the empathy, communication, and care that make it meaningful.
These are design challenges. Becoming smarter by design means acknowledging where technology can help and where human judgment must stay at the center.
Automation as a Pillar of the Modern Learning Organization
For firms aspiring to become modern learning organizations, automation will be one of several essential pillars. Alongside governance, curation, on-demand access, AI, and other capabilities, automation helps create the foundation for a new way of learning. Automation can take care of predictable L&D work, reduce administrative burdens, and ensure essential requirements don’t fall through the cracks. In doing so, it frees up the energy and attention needed for deeper growth. And by handling the routine work in the background, automation gives L&D leaders and people managers more space to do what they do best: teaching, mentoring, and coaching.
But automation alone will not get us there. The firms that thrive will be those that balance automation with culture—using technology to provide structure while keeping space for judgment, empathy, and context. They’ll let algorithms suggest, but let humans contextualize. They’ll recognize that automation can route knowledge, but only people can give it meaning.
As one participant in our private beta group put it: “I love me some automation, but I’m dubious about our firm’s ability to articulate automation rules in a way that won’t create more problems than they solve.”
That tension—between the possibilities of automation and the limits of what can be automated—is precisely the space we are learning into together. It’s one of the many shifts—technological, cultural, human, and procedural—that firms will need to navigate as knowledge management, learning and development, and AI converge.
Working Out Loud
I share all this not as a polished conclusion, but as part of our ongoing exploration. We’re learning together with our beta cohort about what should be automated, what should be left human, and what lies in that middle zone of possibility.
The future of learning and development in AEC will balance the possibilities of automation with the timeless necessity of human connection.
In the end, the goal isn’t automation for its own sake. The goal is helping people grow, projects succeed, and firms thrive. You can’t automate care—but you can automate with care.
This is the journey we’ll be pursuing, as a community, in the months and years ahead: exploring the possibilities and the limitations of automation in AEC L&D, and getting smarter together as we go.
Join the Conversation
I’ve shared these reflections as part of working out loud, but I’d love to hear your perspective. How are you thinking about automation—and its limits—in your firm’s learning journey?
Where do you see automation genuinely helping, and where do you believe human connection is critical?
Send me a note at cparsons@knowledge-architecture.com.