AI, Materials, and the Future of Specification
AI is everywhere right now. New tools. New headlines. Big promises about what it could do. But most of the conversation still lives in theory.
What doesn’t get talked about enough is what happens when you actually try to use it. Inside a real company. With real constraints. Real products. Real complexity.
In this episode of The Trend Report, I sat down with Karli Slocum of 3form and Cosmo Kramer of Bitreel to unpack what it actually looks like when AI moves beyond the buzzwords and into the workflow.
What 3form set out to do had nothing to do with AI at the start. It was a much more practical problem. How do you help designers accurately see, understand, and trust complex materials at scale before they are ever installed? Because when there is a gap between expectation and what gets built, it creates friction and delays. When you’re working with layered materials, embedded textures, shifting light conditions, and nearly endless combinations, a static image doesn’t cut it. Even a standard configurator falls short.
What stood out right away is that this wasn’t a conversation about chasing technology. It was about solving problems. So that’s where they started. Not with AI. With a problem. Because anyone in this industry has experienced the gap between a small sample and a full installation. What looks right in a four-by-four doesn’t always translate at scale. But what’s interesting is how quickly that initial problem led to something bigger.
What Cosmo and his team built for 3form is a browser-based visualization tool that doesn’t just approximate the material. It reconstructs it. Layer by layer. Digitally. The same way it’s built physically. That detail matters. Because once you start digitizing products, you start to run into another challenge. Not visual complexity, but informational complexity.
And 3form has a lot of it. Hundreds of materials. Thousands of variations. Millions of possible design combinations. And on top of that, years of internal knowledge. Product rules. Custom applications. Installation nuances. The kind of information that doesn’t always live in one place. Sometimes it lives in a spec sheet, sometimes it lives in a PDF, and sometimes it lives in someone’s head.
That’s where the AI conversation started to make sense. Not as a replacement for people, but as a way to capture and organize what already exists. Cosmo described it as building a knowledge agent. Something that can pull from structured data, unstructured documents, and even internal conversations. Then use that to answer questions with context and accuracy.
But the key word there is accuracy. Because one of the biggest barriers to AI adoption in this industry is trust. If the answer isn’t right, it’s not useful. And in some cases, it can create bigger problems than it solves. So instead of relying on a generic model, they built in guardrails. Rules based on how products are made. What combinations are possible. What isn’t. What meets code. What doesn’t. Millions of checks running in the background before an answer ever reaches the user. And if something doesn’t line up, it doesn’t just guess. It flags it.
That might be one of the more interesting parts of this entire conversation, because the system isn’t just answering questions. It’s also identifying inconsistencies. If two product documents say different things, it surfaces that. If information is outdated, it brings it forward. It creates a feedback loop that improves the data over time. So, it’s not just a tool for access. It’s a tool for alignment.
From there, the application for AI becomes pretty clear. Internally, it shortens the time it takes for teams to get answers. Especially when those answers aren’t part of the day-to-day. Instead of tracking down the one person who knows, the information is available immediately. And more importantly, it’s consistent.
That’s another challenge we don’t always talk about. Two people can give two different answers to the same question. Not because they’re wrong, but because the information is complex. This reduces that variability. Externally, the opportunity is even bigger because now you’re not just helping your team, you’re helping your customer.
Designers don’t have to wait for a response. They don’t have to dig through documentation. They can ask a question and get an answer in real time. Whether that’s about applications, requirements, or performance. And that changes expectations.
In a world where speed already matters, access becomes a differentiator. If one brand can provide clear, accurate answers instantly, and another takes a day or two, that gap starts to influence decisions. And, we’re already seeing that shift.
Another piece of the conversation that stuck with me was how they’re capturing tribal knowledge. The things that aren’t fully documented. The insights that come from experience. The edge cases that only show up on custom projects.
Instead of trying to force all of that into static documentation, they’re using the AI to interview team members. Pulling that knowledge out through conversation and turning it into something usable.
That’s a big deal. Every company in this industry has knowledge that walks out the door when people leave. This is a way to keep it. And scale it. At the same time, what I appreciated is that they didn’t position any of this as finished. It’s still being tested. Still being refined. Still evolving. I think that’s important, because it reinforces the idea that this isn’t about getting it perfect. It’s about getting it started. Learning as you go. Adjusting based on what works.
That mindset of continuous improvement showed up again when Karli talked about leadership, and the reason this got approved wasn’t because it was AI and AI is the new flashy toy. It was because it solved a problem. That’s it. Not hype, not pressure, not “everyone else is doing it.” A clear problem, a clear outcome, and a path to get there. That’s what made the difference. That might be the biggest takeaway from this entire conversation. Because right now, it’s easy to get caught up in what AI could do. But the real opportunity is in what it should do, where it actually adds value, where it removes friction, and where it makes the business better.
That’s what 3form is exploring. Not just how to use AI, but how to use it in a way that matters. Because at the end of the day, the technology itself isn’t the differentiator. How you apply it is.