🗓️ Nov - Dec 2025 Challenge - Prompt Power!

Hi everyone!

AI isn’t built into Bluebeam Revu (yet 🤩), but that hasn’t stopped users from finding creative ways to use it! This November - December challenge, Prompt Power, is all about how you’re using tools like ChatGPT, Claude, or Gemini to enhance your workflows around Revu.

If you’re part of the Bluebeam MCP Beta, you’ve already seen how AI can enhance your work and we’d love to hear about any standout moments or insights you’ve had. For everyone else, there are still plenty of ways to use AI to boost productivity and streamline workflows.

Maybe you’ve used AI to help analyze data exported from Revu, write Excel formulas for reports, or brainstorm new process ideas. However you use it, we want to see how AI prompts and generative tools are helping you work smarter, stay organized, and speed up your projects.

How are you putting AI to work alongside Revu?

📝 How to Participate

🤝 Share and discuss how you’ve used AI to assist or improve your Bluebeam workflows:

What prompt did you use, or a description of it?
What problem were you solving?
How did the AI output help you work faster, stay organized, and/or approach Revu in a new way?

It could be as simple as prompting ChatGPT to help organize project data in Excel exported from Revu, or as advanced as generating custom automation in JavaScript!

If you'd like to take part in the MCP Beta, feel free to sign up.

Bluebeam MCP Beta

🏅 Rewards

All participants will earn a unique challenge badge and points to level up in the community!

📚 Learning Resources

https://www.youtube.com/%40BluebeamInc

Bluebeam University

Let’s unlock the full potential of AI-powered creativity. Share your best prompts, clever use cases, and AI assist moments below!

Find more posts tagged with

Revu

Community Activity

Challenge

Comments

Josh Gebelein

I have not yet used any AI tool that can reliably read drawings. They don't yet know how to tell where the titleblock is or what is in it, whether it is filled out properly. AI usually fails at taking the logical steps of reading plans and following schedules or details the way a human does (it's hard for most humans too.). This will eventually change I'm sure, but we're off to a slow start. It doesn't help that the way Revit/AutoCAD/Civil3D/etc "print" text and graphics to pdf often result in disjointed digital artefacts where each line of text or detail numbers and sheet names are not connected together and everything still needs OCR'd to be legible. We will probably have better luck using AI directly with digital models than pdf content for drawings, which is currently outside the realm of Bluebeam.

I'm interested in hearing how the Bluebeam AI works, but have very low expectations and there are already so many other offerings in this space honestly hard to imagine what it can do that is 'new'. I fully expect it to fail at most any task apart from doing an advanced search that we could probably do with Copilot or something else just as quickly. Hope I'm wrong, but so few details have been shared that it's hard to tell.

Of course using AI to digest and search normal text pdf documents is easy, we use that all the time. It's basic now. But unless someone says that Bluebeam AI will be smart enough to do things like find all the section headings in a spec book and create bookmarks, or tell me if all the titleblocks and sheet names match the sheet index; I'm going to be disappointed and pan it as deficient. It needs to actually be useful, and provide new capabilities and timesavings.

If anyone at Bluebeam is listening they can reach out to me for ideas. We can tell you what our thousands of Bluebeam users at my company would like to see.

Vince French

I've mainly used AI to improve on formulas in Excel but the most important thing to remember is that you need to know what the answer should be otherwise things could go disastrously wrong.
Even if I type in that I just want the formula improved but the results need to be the same, I end up getting different answers. Run the model again and you can come up with a different wrong answer.
When it does work though, it's great!
Claude with Bluebeam does look very interesting though - summarising large documents, finding details, etc. will all be very handy.

David Cutler

I haven't put my toes directly into the AI waters yet, but I was thinking earlier that perhaps Claude could potential provide a solution to the "split areas" request.

Idea would be to create a markup called "cut line" and place it across an area measurement.

Prompt would be something along the lines of "create 2 new area measurements from the existing area measurement using the existing control points while splitting the the area at the "cut line" markup so that the resulting 2 markups added together equal the original markup"

No idea if it would work, but it's fun to dream!

😎

JonahNelson

I tend to agree with Josh below, Ai has a long way to go to match human intuition and general thought and accuracy. The fact that my comment is only the fourth one after over a week of posting this article also says something. People are skeptical and standing back and waiting to be genuinely impressed. I think there is tons of potential, but wow it has a long way to go especially in an industry where wrong answers can cost easily $1000's of dollars and worse case millions.

Matt Flanagan

Hello Bluebeam Community! 👋

For the November 2025 Prompt Power challenge, I’m sharing a custom GPT I built called Civil RevuGPT. It’s a transportation issue log assistant that works with Bluebeam Revu to turn a Markups List export into a Project Manager-ready transportation issue log and generate a draft email assigning action items to the appropriate staff. A simple Python script then takes Civil RevuGPT’s output and uses Outlook to create pre-populated draft emails that the user can quickly review and send to team members.

🧠 Workflow

Revu Export → Civil RevuGPT → Python Automation → Email Action Items

Civil RevuGPT takes the Bluebeam Revu Markups List and:

Cleans and standardizes comment text
Adds Discipline, Category, Priority, and Location
Outputs a structured table (issue log) that’s easy to filter and assign
Drafts summary email content with bullet-point action items

From there, a simple Python script:

Takes the issue log and email content output from Civil RevuGPT
Groups issues by responsible person
Uses Outlook to create pre-populated draft emails from Civil RevuGPT’s output so the user can quickly review and send action items to project team members

That makes it much easier to go from “we marked it up in Revu” to “someone owns this item,” and then quickly send out action items for the team to tackle.

🚀 Impact

Speed: ~5–10 minutes instead of 45–60 minutes of manual Excel cleanup
Organization: Clear, consistent, filterable issue logs across sheets and submittals

📎 Attached is a one-page overview PDF that summarizes the concept and workflow.

Civil RevuGPT_Prompt Power_Nov 2025.pdf

👤 About me

My name is Matt Flanagan, and I’m a transportation engineer based in Madison, Wisconsin, with expertise in transportation economics, traffic safety, traffic operations analysis, and practical AI applications. I studied physics as an undergraduate and earned a master’s degree in engineering. I’m also an AI enthusiast and have been a near-daily ChatGPT user since 2022.

When I’m not working, you can usually find me grilling, biking, snowboarding, or adventuring outside.

I’d love to see how others are pairing Revu with AI workflows - please share, and thanks for reading!

David Rekker

Excellent work, Matt! Thanks for inspiring us.

ireagle

Wow! You really sound like a pro at AI. Keep up all the great work!

cgmahoney2

This is compelling. Thanks Matt!

Marie W

Really cool work, Matt! As someone in microbiology, I can see a ton of cross-disciplinary potential here. We deal with our own version of “markup chaos” when tracking sample issues, corrective actions, and QA/QC comments across reports, LIMS exports, and audit notes. A tool that can standardize annotations, categorize them, and auto-draft action-item emails would be incredibly useful in lab environments—especially for CAPA workflows, contamination investigations, and audit prep. It’s exciting to see how something built for civil workflows could inspire similar automation in scientific fields!

Matt Flanagan

Thanks so much, David - I really appreciate it! 🙌

This community has sparked a lot of ideas for me, so I’m glad I could give a little bit back. Excited to see what others are doing with AI + Revu for this challenge!

Matt Flanagan

Much appreciated! Just like everyone else, I’m still learning too - and I’m excited to see where all this AI + Revu workflow is headed!

Matt Flanagan

Thanks for the feedback - really appreciate it!

Marco M

Update: Prompt Power Challenge Extended!

Due to the amount of participation and interest in this November challenge, we’re extending Prompt Power through the end of December.

This gives everyone more time to explore AI tools, try out new prompts, and share how you’re using AI to support your workflows around Revu. Your posts so far have been great!

Jump in, experiment, and share your prompts anytime through December 31.

Jim Conley

I use AI to access Bluebeam’s AI features without using AI.

Let me unpack that sentence: One of the ancillary benefits of the MCP server is that it also provides a Revu API that works independently of an LLM. Using Claude Code I’ve written a Python app that pre-processes my takeoffs for HeavyBid.

It’s a game changer.

We have a Heavybid library with about 1400 activities that’s very granular and detailed (i.e. Offsite 100mm Asphalt Base Prep <1000 m2). Heavybid organizes activities under Bid Items and once Bid Items are created you can import activities. If the activity number matches a pre-existing activity in the library it will import the activity into the estimate pre-populated with crew and resources. This makes having a takeoff with Activity Codes very helpful.

Historically, I’ve exported my takeoffs to Excel to prep data for Heavybid. Efforts to build a library of takeoffs in Revu were frustrating because of the lack of API hooks and documentation.

The problems with Tool chests when you’re dealing with 1400 activities are performance, organization, maintenance and consistency. Tried all sorts of approaches but found using a minimal set of takeoff items with enough info to further update when I got it into Excel was the most productive approach. Now I update them in real time from a companion window.

The software polls markups every seconds and shows them in a structured treeview of Bid Items and Activities. When a new markup is detected it appears in a list at the bottom and displays a list of possible matching activities based on the current markup Subject and type. For instance, a Polylength named ‘300’ returns a list of linear measurement activities with ‘300’ in the title.

I can also manually filter the tree which updates with every keypress. Once I select an item it updates custom fields in the markup and changes the color and appearance so it’s easy to see its been processed. I can also drag and drop to update the fields. I can export the list from the app ot generate it from inside Revu.

The activities are pulled from an ODBC data source to an Excel dump of Activities from Revu. CSV sucks as a data source but it makes syncing current Heavybid server data trivial. By skipping the LLM ‘thinking…’ it’s also really fast, real time fast. Had an estimate with about 150 takeoff items last week and it performed admirably, even while demonstrating new voodoo to the guy who signs my paycheck.

So that’s how I’ve been pushing the envelope with the new AI features. My Python sucks but Claude code does a beautiful job of vibe coding based on my feedback and it's a passive way to learn a new language.

Kudos to the dev team - Revu Max is going to be a winner.

Matt Flanagan

Love it, nice work Jim!

Jim Conley

Thank you. Saw what you were up to above and figured I needed to bring my A game if I want to play on this court.

Haven't been this impressed by a product update in a long time. I've been hoping for a proper API for a decade and somebody at Bluebeam made the call to leapfrog hard into an MCP server. The wild part is they opened up the hood for the code monkeys by adding a feature for the end users. As long as they don't price gouge with Max this sort of continual improvement should silence any debates about the value of a subscription model.

We're going to see some absolutely interstellar extensions to Revu in the next six months…