Skip to content 
Revit 2027 doesn’t feel revolutionary. But according to Interscale BIM-CAD Technical Expert David Gerber, that may not be the most important thing about the release.
The practical updates are useful. Rule-based numbering, improved tagging, and other modelling changes can remove repetitive work from everyday Revit workflows. But the bigger shift may be happening around the model itself, as AI and connected data begin to change how people access and use BIM information.
David works at the intersection of BIM, CAD, automation, and AEC workflows, helping project teams implement Revit and related technologies. That gives him a practical perspective on new software releases: what actually changes the way a project gets delivered?
For Revit 2027, his answer is twofold. Rule-based numbering is the feature teams may feel first. AI and connected model data could have the bigger impact over time, but only if firms have the standards, model health, and workflows to support them.
We spoke with David about which new Revit features can realistically change today, where AI still needs caution, and what AEC firms should consider before upgrading.
Rule-based Numbering is the Practical Standout
Rule-based numbering is the Revit 2027 feature David expects teams to feel first. AI may shape the longer-term story, but the numbering tool solves a job many BIM teams already do every day.
Revit 2027 introduces AI, automation, performance improvements and deeper connections across Autodesk’s ecosystem. Which change matters most to you?
David: For me, rule-based numbering is the standout. Because it handles straightforward numbering and tagging without a script, teams can put it to work quickly.
AI has had more attention, although Autodesk Assistant is still in Tech Preview. When I tried it, it felt slow. For some tasks, Dynamo got me there faster.
“What I find more interesting is the ability to connect tools such as Claude to Revit 2027 through Autodesk’s Revit Public MCP Server.”
— David, Interscale
Autodesk’s current server gives AI tools read-only access to model data. Some community-built options are already exploring write and create capabilities, so the direction is moving quickly even though it is still early.
Does Revit 2027 feel incremental, or does it point to a more meaningful shift in BIM?
David: If you look at modelling alone, Revit 2027 feels fairly incremental. Walls hosted on other walls, better tagging, and rule-based numbering are useful. I wouldn’t call them revolutionary.
The larger shift sits underneath those features. As Autodesk opens Revit to AI tools and includes more Forma capabilities in the subscription, the product starts to sit within a more connected working environment.
That could reduce how often people need to jump between applications, tabs, or monitors to find information. The benefit becomes quite practical on site, where a second screen is not always available.
AI Could Open the Model to More People
AI could change who is able to get useful information from a Revit model. Because that information currently depends on specialist skills, many straightforward questions still have to pass through an experienced Revit user.
Where could AI genuinely change how BIM professionals work?
David: For me, the opportunity is less about raw speed and more about who gets to ask the model questions.
If someone wants to know how many doors and windows are on the ground floor, they currently need to understand schedules and filters. They also need to know whether the result is accurate.
If an AI tool can answer the same question in plain English, more people across the project can use the information without becoming Revit experts first.
Revit 2027 starts to make that access more realistic, even though the available AI workflows still have limits.
Where should teams remain cautious, and what does that mean for the BIM Manager’s role?
David: I’ve tested Claude on code-checking tasks, and it performed better than some people might expect.
Although it can review a model and point you in the right direction, I still wouldn’t accept its word that everything is compliant. Use it as a second pair of eyes, not the final sign-off.
The answer is also only as good as the information the tool can reach. A controlled PDF repository may improve the result because it gives the tool access to the relevant codes.
However, the dataset still has to be complete and current. Since many codes sit behind paywalls, a tool that relies on free links may miss something important.
That is why I don’t see AI removing the BIM Manager’s role. It can take some of the repetitive work away, especially when the BIM Manager does not already have scripts. Someone still has to understand the model, set the rules and validate the result.
Automation Only Works When the Groundwork Does
Automation has the greatest effect when it suits the project stage and the underlying data is consistent. Without that groundwork, even a capable tool will struggle to produce a reliable result.
Will automation have a greater day-to-day effect than AI?
David: It depends on where the project is. During concept design, I haven’t seen automation or AI do much inside Revit unless someone has already set up the groundwork.
Forma Site Design and Forma Building Design are more useful at that point because teams can test concepts before taking them into Revit.
They can look at factors such as wind, solar and noise while the design is still taking shape.
That is a useful early-stage workflow, although it is not necessarily an AI story. Once the project moves into tagging, checking and auditing, automation starts doing more of the heavy lifting.
Revit 2027’s rule-based tools suit that production stage because they remove repetitive work from tasks the team already understands.
What stops organisations from automating more of their BIM processes?
David: Standards are usually a bigger obstacle than the technology. The tools are available, but they rely on consistent naming and parameters.
If every project labels information differently, there isn’t much you can automate with confidence.
The problem often grows under deadline pressure. A team creates a workaround to reach one milestone, then nobody has time to fix it properly.
Because the workaround survives, it follows the team into the next milestone or even the next project.
That is often where poor model health begins. It is also why Revit 2027 makes standards more important.
A numbering rule only works when the underlying information means the same thing across projects.
As AI tools become more capable, messy data creates an even larger risk because the answer can sound convincing while resting on inconsistent information.
Connected Data Matters More Than One Application
Connected data matters more than any single application because project teams work across several tools. The quality of the handover therefore becomes part of the model’s practical value.
Is the future of BIM centred on individual applications or connected data?
David: I don’t think the future comes down to one application winning. Plenty of firms still use AutoCAD or Archicad alongside Revit.
What matters is how well those tools exchange information with the rest of the project environment.
A new version of Revit makes Autodesk’s direction clearer through its connections with Forma and emerging AI tools.
If a great modelling application produces a messy handover, the time saved during modelling can disappear later. A more ordinary tool may serve the project better when it exchanges information cleanly.
A connected cloud environment can improve that exchange because everyone works from current information. Teams spend less time searching through old linked files or checking whether somebody has the wrong version.
A common data environment can also keep revisions, issues and RFIs together, which makes ownership easier to follow.
Those benefits still come with conditions. Australian firms should check where the environment is hosted, how the data is governed and when maintenance occurs.
Cost matters as well. To judge it fairly, compare the platform cost with the hours already being lost to file chasing, version confusion and decisions buried in messages.
Revit 2027 includes performance improvements. How much of a real-world performance problem comes from the software, and how much comes from the model?
David: Most performance problems I see begin with the model. Revit 2027’s graphics and performance improvements can help, although overloaded families will still slow a project down. The same applies to very large linked files and years of unresolved warnings.
A well-kept model on an older computer can outperform a messy model on newer hardware. For me, a healthy model has controlled warnings. It also uses consistent families and parameters, with sensible links.
As AI becomes more involved, model health has another layer. The information has to be structured so that the tool can read it correctly.
A model may look tidy on screen while inconsistent data makes it difficult for an AI tool to interpret.
AI can identify some red flags, but it cannot make a poorly structured model healthy by itself.
An Upgrade Has to Suit the Way Your Team Works
A Revit 2027 upgrade is worthwhile when the release supports work your team actually delivers.
That decision depends on the firm’s templates, standards and support position as much as it depends on new features.
What should an AEC business evaluate before moving to Revit 2027?
David: Start with your existing templates, standards and workflows. Many Revit 2027 features deliver value only when the model is already structured consistently.
If that structure is missing, you can complete the upgrade without seeing much improvement.
Then ask whether the changes solve work your team genuinely performs. Better tagging or new wall behaviour may be useful for one practice and optional for another.
At the time of writing, Autodesk’s support window covers Revit 2027 through Revit 2024. A firm can remain on an older release, although moving outside that window changes the support available when a serious model issue appears.
The biggest mistake would be expecting Autodesk Assistant to do work it is not ready for. Because it remains a Tech Preview, teams should test its supported tasks in a controlled setting before using it on a live project.
Otherwise, you may spend longer forcing the AI workflow than it would take to build a schedule or use a short Dynamo script.
It is also worth checking how well the team uses the features it already owns. If current workflows have not been reviewed, Revit 2027 may simply carry the same process into a newer version.
The decision should begin with the work, the model and the standards. The feature list comes after that.
The Next Shift Will Happen Around the Model
The next BIM shift is likely to change the work around the model before it changes core modelling.
As AI joins existing digital workflows, administrative and information-heavy tasks are likely to move first.
What do you expect to change about BIM over the next three to five years?
David: I don’t expect core BIM modelling to change as dramatically as some people do. The industry has already made the major move from 2D drafting and paper-based processes into digital models.
The next shift is more likely to come from integrating AI into those digital workflows properly.
McKinsey estimates that AI and other automation technologies could unlock about US$228 billion in annual value for the US AEC industry by 2030. It also estimates that AI could automate 50 per cent of non-physical work in architecture and engineering.
Because much of that work sits around the model, I would expect change to appear there first. RFIs and document control are obvious starting points. Contract review, drawing mark-ups, email administration and meeting notes may also change early.
The modelling tool may look familiar for a while, even as the work surrounding it changes substantially.
Which skills will BIM professionals need, and what should AEC firms do now?
David: BIM professionals will need a deeper understanding of how Revit works. When you are directing automation or prompting an AI tool, it becomes more important to know why something behaves as it does. You also need to know how to adjust it when the result is wrong.
Prompting is not simply typing a question and accepting the answer. You need enough Revit knowledge to ask precisely and recognise when the tool has misunderstood the model.
Sometimes the issue sits in how the information is presented rather than in the prompt itself.
I saw that with a title block I built. A person read the information from left to right, while the AI tool read it vertically and missed what I needed.
I made a small change that was barely noticeable to a person. Once the structure was clearer, the tool could extract the data and support the automated workflow.
My advice is to get your standards and model health in order before getting too excited about Revit 2027.
Keep upskilling as well, because knowledge from ten years ago may no longer reflect the best way to work. A familiar workflow can be uncomfortable to change, but good training usually saves time, money, and stress later.
***
David’s view of Revit 2027 comes back to readiness. If a firm has clean standards, healthy models, and people who can challenge an automated result, the release can remove repetitive work while opening new ways to use model data. Where that foundation is missing, fixing it is the first job.
