Highlights of Experimentation Q1 CCC Days 2025

Becky PassnerApril 10, 2025

One way we incorporate our values of Character, Collaboration, and Craftsmanship into our company practices is through our quarterly CCC Days events (named for our core values). During CCC Days, our entire company meets at our Apex, NC, office to work on experiments. These experiments usually explore new features or technologies that might benefit our partners or simplify our workflows. We are excited to share a few of our experiments from Q1 and their possible implications. 

RoleModelGPT

As we continue to evaluate our use of AI and Large Language Models (LLMs) in our daily work, we set up this experiment to look at the process for deploying open-source models in our own environment. This would enable more advanced use cases, such as writing proposals or analyzing business documents with the context of other private RoleModel information. This would help to eliminate biases common to public LLMs, protect our intellectual property, and ensure we are not accidentally exposing customer information to the world. This could also lead to a product offering for our partners. While we didn’t get far on our stated goal, we learned a lot more about the process and uncovered some additional experiments to explore around leveraging ways we could smaller-focused LLMs in the software systems we built for our partners, such as context-aware search.

undefined

Augmented Reality (AR) Construction Measuring

The construction and engineering industries face persistent challenges in translating real-world structures into accurate digital models, a process that traditionally demands significant expertise and time-intensive manual work. Accurate measurements of existing structures are essential to understanding the real-world context of where something will be installed. This experiment explored an innovation to our LightningCAD® framework through an Augmented Reality (AR) measuring and design mode. Rather than just using AR to orient a design for visualization, the user would design the system in AR mode, marking where features would be placed in the real world. This includes extracting 3D details from the environment and placing a railing system from our Railing Designer platform. Our experiment gave us some positive data and more information about what processing would be needed to make this production ready.

LightningCAD Isometric Drawings 

Isometric drawing is a rendering technique drawn from an angle that renders X, Y, and Z axes equally. This is especially useful in construction to visualize how components fit together. In this experiment, we explored some libraries and algorithms to incorporate isometric drawings into an existing project. Our results were promising, with a good base of functionality to continue working on this for customers that need these kinds of rendering for workshop drawings or installation guides. Particularly need to finalize the integration with our PDF library to support exportable drawings.

Hotwire Native

Over the last several years, we’ve embraced the Hotwire style of development within our Rails applications. This has also included several hybrid native applications using the Turbo iOS and Turbo Android frameworks as a thin native wrapper around our mostly web application for a great mobile experience with minimal development effort. With the recent release of Hotwire Native to replace these libraries, we wanted to experiment with upgrading our existing applications and implementing the customizations we’ve seen as common needs. We made good progress with one of our cross-platform applications and have now released these updates to production; and are ready to take on the next project using this approach.

System Monitoring Signal vs Noise

Our team relies heavily on Slack as our primary communication channel, utilizing various channels per project to ensure all critical messages are captured in one place. Over time, the number of messages that are noise from our error reporting or logging systems has grown, and the goal of this effort was to find ways to reduce that noise so the signal (important messages) doesn’t get lost or ignored. We did this by implementing filtering rules and a Slack bot to better summarize information about errors at regular intervals.

It was a great CCC Days, and we are excited to see how these experiments help us further our craft and serve our partners by providing new ways to grow their competitive advantage.