![](images/aec-munich.png) # ifc-data-horse 🐴 > **"There are no wheels yet, but the thing is running"** ## [TL;DR](https://en.wikipedia.org/wiki/TL%3BDR) **This project explores [IFC5](https://github.com/buildingSMART/IFC5-development) and [Native IFC collaborative worklows](https://github.com/buildingSMART/IFC5-development/issues/2) based on Git.** ## Meet the Team * **[Milovann Yanatchkov](https://rvba.fr/my_en.html)** : bug-hunter, aec-free-rider, associate-professor, kebab-case-lover. * **Artem Igumnov** : civil-engineer, aec-ice-breaker ## The Challenge * [AEC/O Hackathon 2025 Munich Edition](https://www.tum-venture-labs.de/events/aeco-hackathon-2025/) * [Problem statement](https://github.com/vyzn-tech/ifc-data-bus?tab=readme-ov-file#problem-statement) ## Problem Statement * Architects and planners rely on a **broad ecosystem of software tools**, * yet **poor data integration between these tools creates major bottlenecks** in the design process. * This fragmentation prevents holistic project optimization and slows down coordination. ### Data Stream * Although **OpenBIM** ✅ aims to support **interoperability**, it still lacks robust mechanisms for > **bi-directional data exchange** ✅ * —particularly when it comes to **re-integrating data into CAD and CDE environments**. * **Existing initiatives often do not prioritize full OpenBIM compliance**, leaving a clear opportunity for alternative, standards-aligned solutions. ### 'Flat Modules' vs 'Monolithic Blob' > At the same time, the emerging **modularized and layerized IFC5** standard ✅ * introduces new capabilities for more flexible, **fine-grained**, and machine-friendly data handling. * These advances open the door to **new workflows** built around decentralized, **event-driven data exchange** rather than **monolithic file transfers**. ## Previously ... ### Zurich 2024 * AEC hackathon Zurich 2024 : **[GitAec](https://gitaec.org)** as a **[Git-native CDE](https://gitaec.org/rvba/docs/wiki/Zurich-hackathon)** platform for AEC. ### Zurich 2025 : Tech Talks * AEC hackathon Zurich 2025 : **Flow-based modeling** discused during **[tech talks](https://www.youtube.com/watch?v=vqkKc3TCWdg)** (youtube) ![](images/gitaec_zurich_2025.png) ### Zurich 2025 : Hackathon ### [vyzn](https://www.vyzn.tech/en) : data-streaming-bim-platform * Joining Vyzn team to work on ifc-data-bus **[Blender client](https://github.com/vyzn-tech/ifc-data-bus/tree/main/2025-01%20AEC%20Hackathon%20Zurich/poc/blender-client)** ![](images/vyzn.jpeg) ### ifc-data-bus 🚐 ![](images/ifc-data-bus.png) ## The Experiment ### Embeding IFC5 Viewer within GitAec * Sources can be find in the [ifcx branch](https://codeberg.org/rvba/forgejo/src/branch/ifcx) of GitAec * The viewer is based on IFC5 BuildingSmart experiment [with some modifications](https://gitaec.org/rvba/ifc5) **First experiments went a bit off the rail...** ![](images/gitifcx.png) **...but eventually, it ended fitting in** ![](images/gitifcx2.png) ## Generating IFC5 * [First test](https://gitaec.org/rvba/ifc-data-horse/commit/3c36ade1140461206173999a88f89a80b9fa72be) from [scratch](https://gitaec.org/rvba/ifc5/src/branch/main/viewer/src/utils/python/hack.py) * [First Mesh](https://gitaec.org/rvba/ifc-data-horse/src/commit/12ed3ee4a136d6350775e4fb365a678d1e833ca5/plane.ifcx) * [Obj exporter](https://gitaec.org/rvba/ifc5/src/commit/931817bb3af368aae7747af2592ef63c461155b0/viewer/src/utils/python/hack.py#L77) ## But there's more ! * **Chatting with a bot based on [Nanobot](https://nanobot.ai) and a [Forgejo-MCP server](https://codeberg.org/goern/forgejo-mcp) and let 'it' produce a cube directly on the Forge on the behalf of the 'user'...** ![](images/aec-bot-1.png) ![](images/aec-bot-2.png) ![](images/aec-bot-3.png) ![](images/aec-bot-4.png) Well .... [not really working](https://gitaec.org/rvba/ifc-data-horse/commit/0dcf8710dd86cb4c4fcb6d01429b1099d5b507fa) But wait, ![](images/aec-bot-5.png) ![](images/aec-bot-6.png) ![](images/aec-bot-7.png) Congrats little bot!