• Welcome to The Building Code Forum

    Your premier resource for building code knowledge.

    This forum remains free to the public thanks to the generous support of our Sawhorse Members and Corporate Sponsors. Their contributions help keep this community thriving and accessible.

    Want enhanced access to expert discussions and exclusive features? Learn more about the benefits here.

    Ready to upgrade? Log in and upgrade now.

And So It Begins

jar546

jar546

CBO
Joined
Oct 16, 2009
Messages
13,270
Location
Not where I really want to be
After several conferences with visits to AI Plan Check companies at their booths, the official marketing has begun.


I have soooooo many questions. The last time I spoke, the different vendors said that their system was limited in scope and still in the testing and design phase, but marketing was on full blast. This is a conversation we must have.

Does anyone have any direct experience with something like this? More to come.
 
jar546 said:
After several conferences with visits to AI Plan Check companies at their booths, the official marketing has begun.


I have soooooo many questions. The last time I spoke, the different vendors said that their system was limited in scope and still in the testing and design phase, but marketing was on full blast. This is a conversation we must have.

Does anyone have any direct experience with something like this? More to come.
Click to expand...
It's been a while, but the stuff I had seen is garbage....I say when AI can design the building, then it will be able to review the plans....But they are hoping the municipalities will be the beta testers and seed money source and then the private sector will try to run us over with it...
 
steveray said:
It's been a while, but the stuff I had seen is garbage....I say when AI can design the building, then it will be able to review the plans....But they are hoping the municipalities will be the beta testers and seed money source and then the private sector will try to run us over with it...
Click to expand...
In Florida, the private sector is gearing up to run all over the building departments.
 
"receive structured feedback—flagged issues, precise code citations and clear corrective actions—in hours, not weeks or months."

Clear corrective actions is more than many AHJs will do currently. Of the many reasons not to provide a clear corrective action is that there are often times more than one possibility.... sometimes there are several possible corrective actions for a given situation. Will the AI provide all of the possibilities? What happens if the corrective action that was provided is less attractive to the owner than one of the others.... but they didn't know about the other possibilities until it was too late to reverse course.

"Plans are automatically analyzed for compliance with the Florida Building Code (FBC) and National Electric Code (NEC)—with more jurisdictions coming soon."

What about residential code?

AI plan checking to the NEC is no small feat. Replacing a savant ???
 
I’ve been following this space pretty closely and can share some context.

Most of the “AI plan check” platforms out there right now are exactly where you described, limited in scope, still in pilot mode, and heavy on marketing. The hard part isn’t getting AI to read drawings, it’s getting it to consistently check against the actual building code with enough detail to be trusted by AHJs, private providers, owner/developers and design teams.

The link you posted (PlanCheckPro.AI) is our platform, and we built it specifically to move past that proof-of-concept stage. A couple of important differences:
  • Scope – Right now it’s tuned for the Florida Building Code and NEC, not a nationwide rollout.
  • Output – The system produces detailed, code-tied comments with corrective actions, not just general flags.
  • Turnaround – Reviews are completed in hours, not weeks, and right now we’re allowing unlimited reviews for free while the industry gets familiar with it.
This isn’t marketed as a replacement for AHJs. It’s a tool for developers, engineers, architects, GCs, and private providers who want to catch code issues earlier, reduce RFIs/change orders, and move projects through permitting faster.

The conversation you’re calling for is the one we’re trying to have, not “is this coming?” but “how fast will jurisdictions and professionals adopt it?” Because whether people are ready or not, this is the direction the industry is heading.

If you click on the website (https://plancheckpro.ai/request-demo/) you can request access to get a demo. You can give it a ride and test as many projects as you like.
 
jar546 said:
In Florida, the private sector is gearing up to run all over the building departments.
Click to expand...
I wouldn’t frame it as the private sector “run all over” building departments. What’s really happening is that private providers and new tools like AI are stepping in to address the backlog and inefficiencies that have plagued permitting for years.

Florida’s construction market is moving at a pace where waiting weeks or months for a first round of comments just isn’t sustainable. The private sector isn’t here to replace AHJs, we’re here to complement them by providing faster, more detailed QA/QC, so that when a set of plans reaches the building department, it’s cleaner, code-ready, and reduces the workload on reviewers.

Think of it less as a takeover and more as an evolution. AHJs will always have the final authority. But if we can shorten review cycles from weeks to hours, eliminate unnecessary RFIs, and help departments focus on what really matters, that’s a win for everyone, owners, contractors, designers, and cities alike.
 
wesley.foster@pacifi said:
If you click on the website (https://plancheckpro.ai/request-demo/) you can request access to get a demo. You can give it a ride and test as many projects as you like.
Click to expand...

How long is the demo good for? Is this just a video of someone using the software, or is this a functional copy we can download and use for 90 days (or 60 days, or 30 days)? Speaking as an architect and a building official, you have to realize that we have lives to lead. I think a 30-day trial period is an absolute minimum for any trial software, and I think for something like this 90 days is a far more realistic trial period.
 
Yankee Chronicler said:
How long is the demo good for? Is this just a video of someone using the software, or is this a functional copy we can download and use for 90 days (or 60 days, or 30 days)? Speaking as an architect and a building official, you have to realize that we have lives to lead. I think a 30-day trial period is an absolute minimum for any trial software, and I think for something like this 90 days is a far more realistic trial period.
Click to expand...
Great question. This isn’t just a demo video, it’s a fully functional platform that you can log into and use right now. At this stage, we’ve opened it up with no 30-day or 90-day limitation. For the foreseeable future, we’re allowing unlimited project uploads and unlimited reviews at no cost.

The reason is exactly what you pointed out: architects, engineers, and building officials need real time to put a tool like this through its paces. A 30-day trial wouldn’t give you the bandwidth to meaningfully integrate it into your workflow or test it on multiple projects. We’d rather you have the space to actually see the value and give us feedback without feeling rushed.

So the short answer, it’s a live, working system, and you can use it as much as you want right now.
 
wesley.foster@pacifi said:
Great question. This isn’t just a demo video, it’s a fully functional platform that you can log into and use right now. At this stage, we’ve opened it up with no 30-day or 90-day limitation. For the foreseeable future, we’re allowing unlimited project uploads and unlimited reviews at no cost.
Click to expand...

So it's an on-line portal that runs in a browser and requires an Internet connection, not software we can download and use off-line? Has it been tested to work with Firefox? I use Firefox at home and that's what's on my work computer -- and parts of the ICC web site do NOT work with Firefox.
 
Yankee Chronicler said:
So it's an on-line portal that runs in a browser and requires an Internet connection, not software we can download and use off-line? Has it been tested to work with Firefox? I use Firefox at home and that's what's on my work computer -- and parts of the ICC web site do NOT work with Firefox.
Click to expand...
Correct, PlanCheckPro.AI is a cloud-based portal that runs in a browser, so yes, it does require an internet connection. That’s intentional, since the platform is designed to handle large plan sets, generate code-tied comments, and allow teams to log in from anywhere without installs or updates on local machines.

As for browsers: the system has been fully tested in Chrome and Edge, which are our recommended environments. It will run in Firefox, but we’ve seen some occasional rendering quirks similar to what you described with the ICC site. If Firefox is your primary browser, you can still use it, but for the smoothest experience right now, I’d suggest Chrome or Edge.

We’re actively working to expand compatibility so that no one is boxed in by their browser choice. Feedback like this helps us prioritize that.
 
Is this a platform where plans are uploaded, a review is done on those plans, and a correction is returned? If so, my question has always been about the format plans take. Every architect and engineer uses their own style, conventions and methods. How does an AI interpret these various types of plans?
 
Sifu said:
Is this a platform where plans are uploaded, a review is done on those plans, and a correction is returned? If so, my question has always been about the format plans take. Every architect and engineer uses their own style, conventions and methods. How does an AI interpret these various types of plans?
Click to expand...
Yes, the workflow is exactly that: you upload your plans, the system runs a full code check, and you get back a detailed correction report with code citations and corrective actions.

On the question of styles and conventions; that’s been one of the biggest hurdles in this space, and it’s where most early “AI plan check” attempts have fallen short. We built PlanCheckPro.AI to read and interpret plan sets the same way a human reviewer does: by looking at both the drawings and the notes/schedules. It doesn’t rely on one rigid template. Instead, the system has been trained across thousands of real plan sets from different firms, with different line weights, layering systems, title block layouts, and drafting conventions.

That means whether an architect calls out an assembly one way, or an engineer uses a different convention for MEP details, the system isn’t thrown off. It interprets the content and ties it back to the applicable section of the Florida Building Code or NEC.

It’s not “perfect”, just like human reviewers, it can miss something if the plans are unclear. But the intent is to give you a structured, code-tied set of comments that are consistent across projects, regardless of the drafting style of the design team.
 
wesley.foster@pacifi said:
I’ve been following this space pretty closely and can share some context.

Most of the “AI plan check” platforms out there right now are exactly where you described, limited in scope, still in pilot mode, and heavy on marketing. The hard part isn’t getting AI to read drawings, it’s getting it to consistently check against the actual building code with enough detail to be trusted by AHJs, private providers, owner/developers and design teams.

The link you posted (PlanCheckPro.AI) is our platform, and we built it specifically to move past that proof-of-concept stage. A couple of important differences:
  • Scope – Right now it’s tuned for the Florida Building Code and NEC, not a nationwide rollout.
  • Output – The system produces detailed, code-tied comments with corrective actions, not just general flags.
  • Turnaround – Reviews are completed in hours, not weeks, and right now we’re allowing unlimited reviews for free while the industry gets familiar with it.
This isn’t marketed as a replacement for AHJs. It’s a tool for developers, engineers, architects, GCs, and private providers who want to catch code issues earlier, reduce RFIs/change orders, and move projects through permitting faster.

The conversation you’re calling for is the one we’re trying to have, not “is this coming?” but “how fast will jurisdictions and professionals adopt it?” Because whether people are ready or not, this is the direction the industry is heading.

If you click on the website (https://plancheckpro.ai/request-demo/) you can request access to get a demo. You can give it a ride and test as many projects as you like.
Click to expand...
That link does not seem to work, but maybe my security...
 
wesley.foster@pacifi said:
Try this instead. https://plancheckpro.ai/
Click to expand...
Questions / comments:

1. Is there a revenue stream beyond the stated use of checking for code compliance? Will the data gleaned from my plan check be sold to any 3rd parties? For example, can a door manufacturer buy info from you on how many doors are being plan-checked on a project, and be given a contact for a sales call?

2. If I'm using it as an architect to run a code check, can the AI report on how the project DOES comply, not just how it doesn't comply? For example, as an architect I note that some building departments will spend their first plan check cycle stapling a list of standard corrections, without looking closely at the plans. Then we take that standard list and write the response, such as: "one hour demising walls already shown on floor plan with legend referring to standard wall assemblies on sheet A-800". Is there a way that your program can proactively point out this kind of compliance at time of first plan check submittal?

3. Can your AI program also perform a cross-coordination check? For example, can it scan the plumbing plans to find where they say "see civil plans for point of connection", and vice versa, and point out where the cross reference is not valid?
 
Last edited:
Yikes said:
Questions / comments:

1. Is there a revenue stream beyond the stated use of checking for code compliance? Will the data gleaned from my plan check be sold to any 3rd parties? For example, can a door manufacturer buy info from you on how many doors are being proposed on a project, and be given a contact for a sales call?

2. If I'm using it as an architect to run a code check, can the AI report on how the project DOES comply, not just how it doesn't comply? For example, as an architect I noted that some building departments will spend their first plan check cycle stapling a list of standard corrections, without looking closely at the plans. Then we take that standard list and write the response, such as: "one hour demising walls already shown on floor plan with legend referring to standard wall assemblies on sheet A-800". Is there a way that your program can proactively point out this kind of compliance at time of first plan check submittal

3. Can your AI program also perform a cross-coordination check? For example, can it scan the plumbing plans to find where they say "see civil plans for point of connection", and vice versa, and point out where the cross reference is not valid?
Click to expand...
Great set of questions, let me take them one by one.

1. Revenue stream / data privacy
No, the data is not sold to any third parties. We’re not in the business of selling leads to manufacturers or vendors. The project data uploaded stays private between you and your team.

2. Reporting compliance (not just deficiencies)
Yes, that’s actually built into how PlanCheckPro.AI works. It verifies and documents where you do comply, it just does not sent out that information. That way, you just get a laundry list of deficiencies and noncompliance. The goal is exactly what you described: to get ahead of “standard correction lists” by showing, up front, that items are already addressed in the plans. That makes your resubmittals cleaner and reduces back-and-forth with building departments.

3. Cross-coordination checks
That’s one of the areas where AI adds real value beyond human plan reviewers. The system can scan across disciplines (civil, MEP, structural, architectural) and identify conflicts or missing references e.g., where plumbing calls out a connection to civil that doesn’t exist, or when structural and architectural details don’t align. It’s not a replacement for full BIM coordination yet, but it does highlight inconsistencies that typically slip through and cause RFIs later.

So in short: no data selling, yes to compliance verification, and yes to cross-discipline coordination checks.
 
Yikes said:
1. Is there a revenue stream beyond the stated use of checking for code compliance? Will the data gleaned from my plan check be sold to any 3rd parties? For example, can a door manufacturer buy info from you on how many doors are being proposed on a project, and be given a contact for a sales call?
Click to expand...
I have two major issues with AI, and this is one of them. No AI company has been profitable yet. Most lose billions then go bankrupt. Even the ones that survive burn billions on server infrastructure that will need to be replaced in a few years before they even ask a customer for money. AI is really good at two things; harvesting data and regurgitating data. Selling data seems to be one avenue for AI companies to explore, but that opens up a whole can of worms I'm not touching with a 10 foot poll. The whole industry is a bubble filled with so many empty promises and needless integration that until someone, anyone, figures out how to make money off AI by itself, I'm skeptical of any promise any AI dev makes.

My other big issue is how inaccurate AI is in general and/or how standardized everything will need to be. I've used some of the best AI systems commercially available. None of them have been able to accurately look at a plan and tell me code issues. Hell, none of them can accurately answer my code questions without some error showing up. AI can easily be tricked and stumble if things aren't exactly how they expect them to be. Export a PDF in a way the program can't read natively? Now it needs to use image recognition instead of just reading text, and you're probably screwed. "AI hallucination" is a term for a reason.

I'm sure AI will get to the point where it can accurately do plan check. It's not there yet. I'm willing to bet I can fool any currently existing plan check AI model into accepting any real plan I give it regardless of how code-compliant it is.
 
wesley.foster@pacifi said:
Great set of questions, let me take them one by one.

1. Revenue stream / data privacy
No, the data is not sold to any third parties. We’re not in the business of selling leads to manufacturers or vendors. The project data uploaded stays private between you and your team.

2. Reporting compliance (not just deficiencies)
Yes, that’s actually built into how PlanCheckPro.AI works. It verifies and documents where you do comply, it just does not sent out that information. That way, you just get a laundry list of deficiencies and noncompliance. The goal is exactly what you described: to get ahead of “standard correction lists” by showing, up front, that items are already addressed in the plans. That makes your resubmittals cleaner and reduces back-and-forth with building departments.

3. Cross-coordination checks
That’s one of the areas where AI adds real value beyond human plan reviewers. The system can scan across disciplines (civil, MEP, structural, architectural) and identify conflicts or missing references e.g., where plumbing calls out a connection to civil that doesn’t exist, or when structural and architectural details don’t align. It’s not a replacement for full BIM coordination yet, but it does highlight inconsistencies that typically slip through and cause RFIs later.

So in short: no data selling, yes to compliance verification, and yes to cross-discipline coordination checks.
Click to expand...
Two questions for you.

1. What LLM or generative AI are you using as a base?

2. What do you charge for your services?
 
arwat23 said:
I have two major issues with AI, and this is one of them. No AI company has been profitable yet. Most lose billions then go bankrupt. Even the ones that survive burn billions on server infrastructure that will need to be replaced in a few years before they even ask a customer for money. AI is really good at two things; harvesting data and regurgitating data. Selling data seems to be one avenue for AI companies to explore, but that opens up a whole can of worms I'm not touching with a 10 foot poll. The whole industry is a bubble filled with so many empty promises and needless integration that until someone, anyone, figures out how to make money off AI by itself, I'm skeptical of any promise any AI dev makes.

My other big issue is how inaccurate AI is in general and/or how standardized everything will need to be. I've used some of the best AI systems commercially available. None of them have been able to accurately look at a plan and tell me code issues. Hell, none of them can accurately answer my code questions without some error showing up. AI can easily be tricked and stumble if things aren't exactly how they expect them to be. Export a PDF in a way the program can't read natively? Now it needs to use image recognition instead of just reading text, and you're probably screwed. "AI hallucination" is a term for a reason.

I'm sure AI will get to the point where it can accurately do plan check. It's not there yet. I'm willing to bet I can fool any currently existing plan check AI model into accepting any real plan I give it regardless of how code-compliant it is.
Click to expand...
I hear you, and I agree with a lot of what you’re saying. The AI industry has been bloated with hype, VC money chasing “the next big thing,” and plenty of companies burning cash without a business model. That’s real. But lumping every AI application into that bucket misses where some of us are taking a different path.

PlanCheckPro.AI is developed by Pacifica Engineering Services (https://pacificaes.com), a Florida-based engineering firm. One division of our company already performs Plan Review and Inspections directly for building departments. We know firsthand the bottlenecks AHJs, architects, and engineers face in permitting because we live it every day. This tool was built to solve those exact issues, not as a generic tech experiment, but as an extension of services we already provide.

A couple of distinctions:

  • Narrow focus – The platform is purpose-built around the Florida Building Code and NEC. It’s not trying to be a generalist like ChatGPT. Every output is tied directly back to a code section.
  • Practical infrastructure – Training a massive general AI model costs billions. Running a targeted compliance tool does not. That’s why we can put this live, in use, without chasing bubble economics.
  • Accuracy with accountability – You’re right: AI can stumble if you throw messy edge cases at it. But the goal isn’t perfection, it’s consistency and speed. If the AI handles 90% of the heavy lifting in hours, reviewers (AHJs or private providers) can focus on the nuanced 10%. That’s a net gain for the industry.

So yes, skepticism is warranted, but this isn’t vaporware or a cash burn experiment. This is a firm that already does plan review and inspections, applying AI to make the process faster, cleaner, and more consistent. And it’s already in live use.
 
wesley.foster@pacifi said:
2. Reporting compliance (not just deficiencies)
Yes, that’s actually built into how PlanCheckPro.AI works. It verifies and documents where you do comply, it just does not sent out that information. That way, you just get a laundry list of deficiencies and noncompliance. The goal is exactly what you described: to get ahead of “standard correction lists” by showing, up front, that items are already addressed in the plans. That makes your resubmittals cleaner and reduces back-and-forth with building departments.
Click to expand...
Thanks, but I want to clarify.
The way I currently get out in front of standard plan check correction lists is to proactively co-opt them for my own purposes. See this example from Los Angeles Dept. of Bldg. and Safety for multifamily housing: https://dbs.lacity.gov/sites/default/files/efs/forms/pc17/PC.STR.Corr.Lst.18-(Rev.-01-01-2023)--.pdf

In the past, prior to initial plan check submittal, I've proactively printed that correction sheet, crossed out the non-applicable comments, and handwritten next to each comment where the answer could be found on the initial plan check submittal. I upload that pre-responded correction sheet at time of first plan check, effectively saying to the plan checker, "we already did your work for you, so don't waste our time by not looking closely during initial plan check".

Assuming for the moment that LADBS will not embrace any AI plan check program in the near future ( no fault of yours, it will just be due to the typical power plays at city hall), your program could still have value at time of initial plan check by publishing their checklists with the responses ("where to find it") already on them.

In other words, a totally clean set of plans with no correction printouts from your program will not convince a manual plan checker that it complies with code, they'll just be wondering if you missed something; thus it will not save them any time (until the day comes that they fully trust your program). However, a standard correction list already cross-referenced to where the compliance can be found on the plans WILL be more productive within the context of their current bureaucracy.
 
You must log in or register to reply here.
Back
Top