There is a tab you keep open while working in Revit. Sometimes it is a regulation document. Sometimes a planning portal. Sometimes a project brief you need to cross-reference. Often all three, and you toggle between them and your model trying to hold the thread.
That friction sits underneath every design decision that requires reference material. Which is most of them.
Arkyv is building the answer to it. Their framing is an AI workspace for architects: regulations, project documents, and company knowledge accessible where you are designing, not in a separate application. The product runs on web, inside Revit, inside Rhino, and inside Archicad as of last week.
What It Actually Does
You connect your regulation documents, your project files (briefs, specifications, uploaded PDFs from SharePoint or Google Drive) and your BIM model. You ask questions against that combined context: code requirements, room schedule queries, model checks. The answers come with citations pointing back to the source document or clause.
An answer you cannot verify is one you cannot rely on in practice. Arkyv gives you the reference alongside the response.
The agent layer sits on top. Describable workflows that run against the model, the documents, and the rules together. Check exits against building code. Flag doors below minimum clear width. List rooms with incomplete data. Export a pre-issue schedule. Tasks that currently require someone to do them manually, against documents held in a different window, with a checklist in a spreadsheet somewhere else.
Why Archicad Is the Right Next Move
Arkyv's customer base reads as predominantly Scandinavian: LINK Arkitektur, FOJAB, DREEM Arkitekter, Lindberg Stenberg. Not an accident. Archicad has strong roots in Northern Europe and many of the firms already using Arkyv are Archicad firms.
The move is also a signal about what they are building. A product that works only inside Revit is a Revit plugin. One that works across Revit, Rhino, and Archicad is positioning itself as tool-agnostic infrastructure. The knowledge layer stays consistent regardless of which authoring environment you happen to be in.
Three Types of Knowledge
Regulations are external and authoritative. They change on a schedule, they are jurisdictionally specific, and a wrong answer carries professional liability. This is where citations matter most and where generic AI tools perform worst. The hallucination risk is highest where accuracy is most consequential.
Project documents are what the team has connected: briefs, specifications, planning documents uploaded directly or via SharePoint and Google Drive. The query layer lets you ask against these alongside the model rather than holding them in a separate window.
Company knowledge is the most underrated of the three. The scripts, the workflows, the checklists, the way your office checks room schedules before issue. These live in specific people. When those people leave, the knowledge goes with them. Arkyv's reusable workflow layer is a partial answer to that, and it does not get talked about enough.
The Deskilling Question
Whenever a tool automates something an architect currently does manually, the deskilling concern is worth raising. Does the architect who uses Arkyv to check code compliance develop less code knowledge over time?
Arkyv's citation approach makes this less of a concern than usual. The answer comes with a reference to the clause, the document, the source. A junior architect using it is not just getting an answer. They are being shown where to look, which is closer to how a senior colleague would mentor them than to what a black-box AI does.
Whether that holds depends on whether users follow the citations or just accept the output. The tool provides the scaffolding. The professional habit is still theirs to develop.