Architects at Slantis Build Their Own Revit Tools. Here's What That Looks Like.

In September, Mercedes Carriquiry and Nicolás Martínez from Slantis walked me through how AI adoption actually spreads across a 250-person architecture studio: the maturity framework, the strategy, the culture work. It was a good conversation, but it was still a conversation about a framework.

Catch up on the first session here.

This time, applied research lead Massimiliano Fraga (Massi) and four others from Slantis's production and research teams got on camera and showed me the tools they built. Actual Revit add-ins, running live, made by architects who aren't developers by training.

A framework tells you adoption is possible. Watching four people demo tools they built in their spare time tells you what it costs, what it looks like day to day, and where it pays off.

Watch session 2 here.

HD Details: cutting a 10-hour task to five minutes

Nazarena (Natcha), a project architect at Slantis, built HD Details with her colleague Bas after a request landed in a Slack channel: could anyone automate pulling construction details out of finished projects and into the firm's shared detail library?

The old process was brutal. Export every detail from a closed project as an individual file, strip out anything tied to that client's standards, rename and tag everything by hand. Natcha said one project used to eat a whole afternoon, and on one roof-detail project she spent about 10 hours.

The tool walks through that same process as a guided sequence inside Revit. Pick the drafting views you want to keep (it remembers which ones you've flagged as you work), dump them into a disposable intermediate file, strip client-specific line styles and keynotes automatically, then tag and export each detail into the library. It now takes about five minutes, and a person is still in the loop at every step, deciding what's good enough to go into a shared resource everyone else will pull from.

Natcha had never built a tool before this one. She started with a Dynamo script a colleague had already begun, then kept extending it and iterating with the help of Claude Code until it became HD Details.

Model Care: letting a BIM manager stop being a full-time janitor

Florencia (Flor), a project architect and BIM specialist at Slantis, built Model Care to solve a different problem: Revit models degrade, and most firms only find out when something is already slow or broken.

Model Care runs a full set of health checks against a model at any project stage and scores it against criteria Slantis weighted themselves. Flor's point was that off-the-shelf model checkers count every flagged item the same, and that isn't how it plays out on a real project. A missing phase assignment doesn't matter the way a broken shared coordinate system does. The dashboard separates "fix this first" from "fix this when you have time," explains why each issue matters, and exports a report the team can hand to a client or pull into Notion (it's my favorite database that runs all of TRXL too), where Slantis already tracks every project.

I asked the obvious question: why not buy one of the many existing health checkers? Her answer was that none of them let Slantis weight the checks to their own workflow or project phase. A generic plugin isn't going to close that gap.

She also showed a companion tool, Revit Detective, a one-click diagnostic for the BIM specialists who want a deeper look. Both came out of production time, not weekend hours. Slantis treats this as sanctioned work, which is a big part of why it exists.

Spawn: turning a photo into a native Revit family

Spawn, built by Ivan (Ivi), is the one that's hardest to believe until you watch it. Ivi is an architecture student who has been writing code on the side for about a year, and he built a Revit add-in in Claude Code that generates native Revit families from a reference photo or a manufacturer's cut sheet.

Drop in a photo of a lamp, pick a model (he used Opus 5.5 for the demo), set the level of detail, and Spawn builds a 3D object. It flags what it couldn't model with confidence (in the demo, the power cord) and lets you refine with plain-language prompts like "remove the smaller shade" before generating the family. Given a spec sheet with real dimensions, it reads the measurements directly, and one experimental mode produces a parametric family you can keep adjusting inside the project.

Ivi was upfront that parts of this are still experimental. He built in hard spending and time caps, plus per-user limits on the back end, because he has read the stories about surprise bills from AI tools with no limits set. If this category of tool spreads, the cost guardrails will matter as much as the output.

Magic Tools: making Revit's buried commands findable

The last demo, from Joaquin (Joacho), hits something every long-time Revit user knows: the ribbon holds more tools than anyone can remember, and most of them stay buried.

Joacho and a small research team (Fede, Maru, and Emmy) built Magic Tools, a pyRevit-based launcher that surfaces the commands you need. An assistant that would answer "is there already a tool for this?" in plain language is still in progress. He showed two once-separate tools, a category manager and a batch filter override, merged into one view for managing view templates after the team realized they were solving the same problem from two directions. He also showed a quick-access panel that changes what it offers depending on whether you're in a sheet, a view, or have an element selected.

Joaquin has no formal coding background. Dynamo and Revit were as close as he had gotten before this year, and he doesn't think of this as becoming a developer. He thinks of it as customizing the tool he already uses every day.

Slantis is open-sourcing a subset of Magic Tools through the pyRevit community, after testing to make sure nothing ships that could crash someone else's model. You can find them at magictools.slantis.com.

What this tells you (Don't skip this part.)

A project architect, a BIM specialist, an architecture student, and a project developer each had a recurring annoyance in their own workflow, and each had dedicated time to go fix it. Slantis calls that time "the branch": the space between projects where people can pursue a skill or build something.

The tools are useful. The structure behind them is what matters more. This studio decided that giving people time to solve their own problems with these new AI tools is worth more than squeezing every hour into billable work. Massi put it plainly on the stream: architects are natural problem-solvers, and when you hand them tools like these, you get more of this.

None of the tools are for sale (yet). Slantis is still running safety checks before anything leaves the studio, apart from the open-source pieces.

If you run a studio and wonder whether this could happen under your own roof, Slantis's answer is to give people the time and see what comes back.

If this sounds interesting, watch the video here.