AI

We have a lot to say about AI.

The beginning parts of a project are the most joyous. Understanding a place. Drawing the first line. Testing an idea. Working through a plan. Thinking spatially. Discovering something you did not expect. Moving between a sketch, a drawing and a model until the building begins to reveal itself. These are not simply tasks to complete more efficiently. They are acts of judgment, imagination, and craft. They are why many people become architects in the first place. Good software should make this process more fluid, expressive, and powerful, not rush to do it for you.

Architecture also contains an extraordinary amount of work that has little to do with designing. Coordinating changes across drawings. Updating schedules. Managing specifications. Renaming files. Tracking revisions. Maintaining libraries. Checking dimensions. Finding product information. Preparing drawing sets. Re-entering information that already exists somewhere else. Much of this work is necessary, but little of it is why someone becomes an architect. The opportunity for better software is enormous: remove the mechanical work surrounding architecture so people can spend more of their time actually practicing it.

This is where AI becomes genuinely exciting. It can find information, recognize patterns, reconcile inconsistencies, classify objects, maintain data, surface conflicts, search enormous libraries, and automate repetitive actions. It can help check a drawing set, research a product, populate a schedule, compare revisions, or find the detail you know exists somewhere. These are jobs where the machine's speed and persistence complement human judgment. The goal should not be to remove the architect from architecture. It should be to remove the tedious work that keeps the architect from architecture.

Creating is not wasted effort on the way to an answer. The effort is part of how we learn to see, think, and exercise judgment. Drawing something teaches you about it. Wrestling with a plan develops intuition. Making a bad decision and correcting it builds expertise. If we routinely outsource that struggle, we risk weakening the very faculties that make someone good at the work. There is a human consequence too. Meaning comes from having agency, developing mastery, and seeing yourself in what you make. Technology should expand those capacities. We should be careful about using it to quietly take them away.

Generative AI makes for a spectacular demonstration. Type a sentence and watch a building appear. That is easy to understand, easy to market and easy to fund. The harder problems in architecture are less theatrical: geometry, interoperability, documentation, coordination, dependencies, performance, and the accumulation of thousands of decisions over years of work. Solving those problems requires deep knowledge of how architecture actually happens. We believe the greater opportunity is not generating the beginning faster. It is building a better foundation for everything that follows, while giving people extraordinary tools to bring their vision to life.

Always stay in control. It's your work. It should have your unique fingerprint.

We believe technology should give architects greater agency, not progressively remove them from their work. Corvus uses AI to take on research, checking, coordination, and repetition. When it comes to the creative process, you should have complete control. Draw every line yourself. Ask AI to draw something for you. Or work somewhere in between. Stay in control of every decision.

Keep the pen. Add the prompt.

Some new tools imagine a future where the prompt replaces the pen. Describe what you want, let AI generate it, then clean up what it made. We think architects deserve something better. Corvus gives you powerful drawing and modeling tools alongside AI that can use those same capabilities. Draw when you want to draw. Prompt when you want help. Move freely between the two. AI expands what you can do without taking away what you already know how to do.

You decide where AI begins.

There is no universal line between meaningful creative work and tedious production work. One architect may love drawing every piece of furniture. Another may want a specification sheet turned into clean 2D and 3D geometry automatically. Both are right. Corvus does not impose a philosophy of practice on you. Use AI extensively, sparingly, or hardly at all. The important boundary is not one we set in the software. It is the one you choose for yourself.

AI should understand what you're working on.

Ask whether a door will hit a light fixture and AI should understand the question in the context of your actual project. It should know the geometry of the door, the dimensions of the fixture, where each sits in space and the manufacturer's information stored in your library. That requires more than language. AI needs access to the drawings, models, objects, relationships and information that make up the building. Only then can it become genuinely useful.

Give AI the work you don't want.

Some tedium is administrative. Research a building code. Find a specification. Check dimensions. Look for inconsistencies. Compare revisions. Some tedium happens right inside the act of drawing and modeling. Populate a site with hundreds of trees. Replace the same element everywhere. Recreate an existing facade from a photograph. Turn a manufacturer's cut sheet into usable geometry. What counts as tedious changes from person to person. Corvus lets you decide what to delegate.

Turn information into something useful.

Drop in a manufacturer's specification sheet and ask Corvus to create the object you need. AI can read the dimensions and product information, use the appropriate drawing and modeling tools, create clean geometry and attach the relevant data. Instead of searching for a questionable model or rebuilding an existing product by hand, start with the source itself. Review the result, refine it if needed and keep designing.

Let AI do the looking.

Buildings contain thousands of decisions, and small conflicts can hide among them. Ask Corvus to look for what you might have missed. A fixture hanging too low for a door. An incompatible electrical requirement. A dimension that does not agree. Information in a schedule that conflicts with the model. AI can continuously examine geometry, project data and source information together, helping you find problems while they are still easy to solve.

Research without leaving your work.

Architecture requires constant investigation. What does the code require here? Which product meets these constraints? Where did we use this detail before? What changed between these revisions? Ask in ordinary language and let AI search the information connected to your work. The answer should arrive with its source and context, so you can verify it and exercise your own judgment. Less hunting through documents and websites. More time thinking about the building.

Born with AI. Not retrofitted for it.

Corvus is being built from scratch in a world where AI already exists. It is not a feature we have to bolt onto decades of software architecture. That changes what becomes possible. AI can work across drawing, modeling, objects, libraries, worksheets and project information because those capabilities are being designed to work together from the beginning. The result is not an AI feature around the edges of your software. It is intelligence available wherever you choose to use it.

Creating is part of thinking.

The work of designing is not wasted effort on the way to an answer. Drawing teaches you about what you are drawing. Wrestling with a plan develops intuition. Trying something, seeing why it fails and trying again builds judgment. These acts shape the architect as surely as they shape the architecture. AI can accelerate extraordinary amounts of work. But faster is not automatically better when the work itself is how we think, learn and discover.

More architect. Less operator.

The promise of AI should not be fewer architects doing less architecture. It should be architects spending less time operating software, hunting for information, repeating themselves and fixing preventable mistakes. Corvus combines powerful human tools with powerful machine capabilities and leaves the relationship between them in your hands. Use AI to remove friction. Use it to extend your capabilities. Use it differently tomorrow than you do today. The architect remains in control.

Corvus is not an AI wrapper.

We are not putting AI on top of someone else’s design software. We are building the design software itself, from the geometric kernel and object model to drawing, modeling, data, collaboration, and presentation. AI will be considered at every layer of this. That means AI can work with the same geometry, objects, relationships, and information you do, rather than trying to interpret them from the outside. As AI becomes more capable, so does Corvus, not because we replace the architect with a better model, but because we give better models a deeper understanding of the architect’s work. Any other approach would be fly-by-night. We want to make this lasting.

Join our waitlist.

You’ve already waited 49 years. What’s another two? AEC professionals have spent decades jumping through hoops and creating workarounds for convoluted software. No one dares to build something that will challenge the status quo. It's time for something fundamentally different and unapologetically human. Join the waitlist.