Experience Report: Why I Now Recommend External Integrations Over Matterbridge
Hi everyone 
After the release of external integrations in Gladys, I wanted to share my experience with Matterbridge and explain why I will now direct users towards external integrations.
When I discovered Matterbridge, I found the concept really excellent: allowing devices that are not yet Matter-compatible to be integrated with Gladys through an abstraction layer. I believed in it, recommended this solution, and even directed several developments in this direction hoping it would allow more users to join the Gladys ecosystem.
After about two months of use and community feedback, I must admit that I am no longer entirely satisfied with this approach. In hindsight, I think that Gladys external integrations are superior in every way, and this will now be the solution I recommend.
Why?
1. A Much More Robust User Experience
Matterbridge is a very good project, but it is primarily designed by developers, for developers.
From the feedback I’ve read on the forum, the experience often becomes complicated when something doesn’t work. Problems are difficult to diagnose, and solutions are not always obvious.
In my opinion, this mainly comes from the technical choices made.
Matterbridge works a bit like Gladys v3: plugins are installed directly in the main container, with dependencies installed at startup (runtime). This approach is convenient at first, but it quickly becomes fragile: a dependency update can break a plugin without warning.
Gladys external integrations follow a very different approach:
- each integration is isolated in its own container;
- dependencies are installed at build time, not at startup;
- one integration cannot break the others;
- updates are much more predictable.
This architectural choice completely changes the game in terms of stability.
2. Matter’s Limitations Remain Matter’s Limitations
Matterbridge remains dependent on the Matter standard.
However, if some devices are still not Matter-compatible today, it’s often not by chance: it’s often because the protocol does not yet allow expressing all their features, or because certain types of devices simply do not exist in the specification.
In other words, Matterbridge will always be limited by the possibilities offered by Matter.
On the other hand, a Gladys external integration can exploit 100% of a device’s capabilities, without being constrained by an intermediate standard.
3. One Less Dependency
Finally, Matterbridge is an external project.
When some users encountered bugs, it was often frustrating for me because I simply couldn’t fix them. I was dependent on the evolution of another project.
With external integrations, we fully control the platform. Developers can publish their integrations freely, evolve them at their own pace, and we can improve the ecosystem without depending on an intermediate layer.
And Now?
Now that external integrations exist, I no longer see any reason to continue promoting Matterbridge for this use case.
External integrations fulfill exactly the same role, with better stability, a better user experience, and many more possibilities.
As external integrations replace the Matterbridge plugins I had recommended, I will update the documentation and gradually remove Matterbridge tutorials in favor of these new integrations.
Regarding the plugin factory, it represents a monthly cost (servers + AI). I don’t think I will keep it in the long term: I prefer to invest these resources directly in the development of Gladys. 
To Conclude
I hope you will understand this change of position. It’s not about being indecisive: AI is evolving our way of developing at an incredible speed, and I think it’s important to know how to question your choices when better solutions appear.
In the end, I mainly retain one thing from this adventure: Matterbridge gave me the idea that led to Gladys external integrations.
Without this experience, this feature would probably never have seen the light of day. And today, I genuinely think it’s one of the most important evolutions of Gladys in a long time.
Thank you all for your feedback, your tests, and your trust 