Preparing your firm for an AI-orchestrated future

Large language models aren’t just sophisticated chatbots anymore. They’re becoming orchestration engines capable of connecting and coordinating your entire data environment. We are moving into an era of context graphs, but more about that later.

First, we need to start preparing now for when AI software becomes valuable for design and BIM processes.

Most AEC firms today operate with disconnected tools. You have separate systems for project management, BIM coordination, document control, communication, and design. We’ve all gotten used to switching between Excel, AutoCAD, Robot, Photoshop, Rhino, Navisworks, PowerPoint, Miro, Word, Revit, and countless other software. We manually transfer context and information between them.

Model Context Protocol (MCP) changes this fundamentally. It allows a large language model to act as a universal interface by reading from your email, pulling data from your CRM, accessing project documentation, and even interacting with your design software.

Personally, I’ve been using Claude Projects with MCP connections across our business systems: Google Calendar, Slack, Notion, HubSpot, Gmail, and Google Drive. Claude pulls contract details, tracks client conversations, analyzes sales pipelines, and prepares board meeting materials by querying connected systems directly. It’s not replacing judgment. It’s eliminating the mechanical work of gathering context before making decisions.

We’ve seen proof-of-concept MCP implementations for both Revit and Rhino. The technology works. 2026 is likely the year this moves from interesting demos to enterprise-grade, production-ready solutions.

One alternative to disconnected tools would be having one tool that does it all: design, documentation, and coordination. We have yet to see any software vendor pull this off. But perhaps there is a way for firms to get close.

AI-assisted coding is opening up the world of developing design and BIM software to more people. Digital technology specialists in AEC firms used to script Dynamo and Grasshopper graphs. Now they can create text-based code. You might ask what’s the big difference. Beyond flexibility and performance, the main benefit of text-based coding is the potential it has to scale across your firm.

What used to be made for a single project can now be used by hundreds of projects if you employ the right expertise to make it so. As AI-assisted coding gets better, the bottleneck isn’t writing code anymore. It’s understanding the domain problem and architecting the right solution. This is perhaps the most significant transformation for toolmakers in our industry, and it’s happening right now, not in some distant future.

Successful AI adoption isn’t primarily about technology. It’s about organizational readiness and leadership. Here’s what I believe actually matters:

I spoke about the architecture firm ATI Project at Autodesk University in 2024. They were transforming their archives of design data into a database they could query and run machine learning algorithms from.

Making data digital and formatting it in a way that’s computable is a key first step. Large language models and machine learning algorithms need to be able to consume your data. It’s not glamorous, but it sets you up for a great foundation.

Then it’s also imperative to have a good data capture process to build context graphs for ongoing projects.

As more people can code, firms should develop the strategic capabilities around how to build tools. It’s not just about creating a proof-of-concept for that one project. It’s about acquiring the skills to build a small design or BIM tool and being able to scale it across the business.

Build your own infrastructure to develop and maintain tools over time. Track usage. Automatically catch errors. These capabilities are great preparation for the world where AI is actually helpful and valuable on design and BIM projects.

Be strategic about your vendors. Ask yourself these questions:

Can you customise on top of their platform? Can you design and scale your own tools as a way of institutionalising the unique knowledge of your firm?

Or are they point solutions that solve a problem but don’t enable you to make the solution fit your process? Do they allow you to build value unique to your firm over time? Do they connect to your other tools?
Choose platforms that you can customise.

In a world of AI, learning is ever more important. It takes time to learn. If you are able to start early and keep a high pace of learning, then it’s really hard for your competitors to catch up with you.

When I presented on MCP at Digital Construction Week London in 2025, it was emerging technology worth watching. Months later, we’re deploying it in production for business management. We’ve seen proof-of-concept implementations for design software. 2026 is when this moves from possibility to practice.

The gap between “interesting experiment” and “industry standard” is shortening. Firms that excel at organizing their digital foundation, building AI literacy, experimenting with integrations, and developing custom capabilities will have massive advantages. Those waiting for turnkey solutions will fall behind.

We feel the shifting winds ourselves at Reope. Our position as “the customization layer for AEC software” takes on new meaning in an AI-orchestrated world. We’re not just some of the best in the world at connecting Revit or Rhino to specific workflows anymore. We’re helping firms build the capabilities they need to stay relevant in a world of AI.

My clear advice is to start now and to start by building capacity. If you’re looking for the one AI software that will solve all your problems and trying 5 new ones each week, you are probably wasting your time.

Having more and more software vendors builds more and more dependencies. This is a business risk to be aware of. You can mitigate it by selecting a few platforms and starting to build capacity to make the design and BIM tools your firm needs on top of those.

As we move into 2026, context graphs are emerging with a promise to become the critical infrastructure layer that makes AI orchestration practical at enterprise scale.

While MCP provides the protocol for providing context to large language models, context graphs have the potential to put it all into perspective. They map the relationships between your projects, people, documents, decisions, and design elements. This creates a data structure that AI can traverse to understand not just what data exists, but how it connects and what it means.

One of the best places to see where this technology is heading and learn from early implementations is conferences like Digital Construction Week. The gap between firms that understand this and those that don’t will define their competitive advantage in the years to come.

Joachim Viktil is an engineer and MBA graduate with nearly 15 years of experience in the AEC Industry. Over the course of his career, he developed a passion for leveraging technology to create a better built environment.

After a more than a decade at an international engineering services firm followed by a brief stint at an AI startup, Joachim joined Reope, where he serves as CEO. Reope spearheads next-generation BIM coordination and custom design software development, partnering with architecture and engineering firms to unlock the full potential of their data. Joachim’s deep industry expertise, coupled with his entrepreneurial spirit, has positioned him as a trusted voice on AI adoption and human-centered innovation in the AEC sector.