Pranav

Amazing to see this problem getting noticed and now that it has gotten VC's interest. Pretty sure there is an investment post coming and can't be excited more. Also, I'm that engineer currently waiting on a Revit model to load and have been doing this everyday for >11yrs. Although I agree with the general premise in need of a better ecosystem, I want to highlight the problem here is not ONLY the software used. Here are my thoughts. 1. "Software built in 1997" is a misleading frame. Windows was built in 1985. What matters is the updates. Revit has had constant feature upgrades (it's not there yet) and Autodesk has built BIM 360/ACC/Forma around it. You can disagree with the pace, but calling it frozen in time is not quite right. 2. There are several tools that assumes browser based, cloud native is the way to go. Cloud Native is right and I like the approach Autodesk has taken here with cloud collaboration models for Revit. Again, these are MASSIVE models, assuming we aren't discussing a residence or an apartment complex, a single facility can have hundreds of thousands of elements/10s of 100s of design files/companies across dozens of disciplines. On one of my projects, atleast 25 different global companies - design partners to low voltage contractor, their drafting contractors worked on the same Revit project at various timezones. Incredibly difficult to cater such a big audience. Sometimes the projects work in silos by necessity, not ignorance. 3. Revit's worksharing capability exists for a reason. "A structural beam moves on Tuesday. The MEP consultant doesn't find out until Friday." NOT TRUE, unless they use outdated practices, if they can't adapt to a feature Autodesk introduced in 2016, I don't expect them to use/adapt to a new tool in 2026. 4. "The platform also doesn't connect the 3D model to the physics-based calculations engineers need to run alongside it. Those happen in separate tools — notably Excel and other third-party software - PARTLY TRUE, although several Revit compatible tools exist, opportunity to create internal tools exist, stamping engineers prefer calculations on tools that they're comfortable with. RISA for structural design calcs, converting it to Revit is tricky. Interoperability challenge - Yes, wish there is a better way. 5. The coordination failures described aren't Revit's fault, they're execution failures. When piping lives in Plant 3D and structures in Revit, yes there WILL be interoperability issues. Surpringly large no. of contractors/trade partners doesn't use Revit in the US, they're stuck on AutoCAD. 6. I worked on some of the most complex projects in the NA, from tallest building on the continent to one-of-a-kind manufacturing facility. I've never seen a single project use all of Revit's capabilities. Why? Because projects trade off features for speed. Owners want highest quality, under budget, fastest timeline and I remember my ex manager said you typically only get two of the three. That's a construction reality, not a software problem. 7. The piece frames this as "Revit bad, AI fixes everything." But the actual bottleneck is: LOD, EIR standards not being enforced, BIM Execution Plans not followed, trade partners using different execution strategies. Organizational discipline without disrupting project schedule is where we will need AI/LLMs. 8. Now on the "AI will automate MEP design" claim. I work alongside MEP engineers daily. Yes, a lot of the work is rule-based. But the rules are layered - building codes, local amendments, AHJ interpretations, equipment vendor specs, owner preferences, vendor availability and site-specific constraints all interact. An AI placing fire alarms in a generic floor plan is a demo. Doing it on a industrial facility or hospital where the mechanical, electrical, and plumbing systems are deeply interdependent? That's where every "automate in minutes" pitch falls apart. 9. The article says document review "catches only 30% of issues" and costs $50-100K and highly manual. That number might be real, but the reason isn't bad software, it's that the reviewers are working from incomplete designs, outdated tools. In cases, Disciplines submit at different completion levels. You're reviewing 60% MEP against 90% structural. AI reading those same incomplete documents won't magically catch what isn't there yet. Garbage in, garbage out. 10. There are already players chipping away at interoperability like IFC.js, Speckle, OpenBIM, Autodesk's own evolving platform. The ecosystem is moving. It's not there yet, its upto the firms to bridge those gaps, democratize access. Compared to 2020, it's much easier to extract the data from these tools today and owners'/GCs must build an intelligence layer on top of these 11. Do I want better tools? Every single day. As someone who scripts automation on top of these platforms daily, I know the pain intimately. But the gap isn't "nobody's thought of this."

You can now enable Claude to use your computer to complete tasks. It opens your apps, navigates your browser, fills in spreadsheets—anything you'd do sitting at your desk. Research preview in Claude Cowork and Claude Code, macOS only.

You can now enable Claude to use your computer to complete tasks. It opens your apps, navigates your browser, fills in spreadsheets—anything you'd do sitting at your desk. Research preview in Claude Cowork and Claude Code, macOS only.