Dev Log

We have settled on a new kind of syncing mechanism, one we have not seen tried before. As far as we can tell, there is no open-source project doing it this way.

One thing I love about AI is that difficult things are still difficult. AI has not made a problem like this easy, but it is helping people take on ideas that once felt out of reach. It is encouraging to see what people can now attempt.

Some people ask why they should buy software if they can create it overnight. But you cannot DIY this kind of software overnight. Even with the most capable and expensive models, it still takes months of engineering, iteration, effort, and energy to build something like this. Maybe in ten years that will change. For now, AI has not turned this into a weekend project, and honestly, I find that encouraging.

We have been trying two different approaches to syncing over the last six months.

The first one worked to some extent, but not well enough. The second one was much less stable. We are now working through a third approach.

Sync is one of those foundations we cannot work around. If it is not reliable, the product has a fundamental problem, so we cannot keep moving forward until we get it working properly. That is also why there have not been many visible product updates recently. Right now, the work is concentrated on making sync stable enough to build on.

Even with all the AI helping us, there are still some things that cannot be done easily. Sync appears to be one of them.

We have been working through the architecture for syncing in RightPlace.

Sync touches everything: core data, Applet databases, project files, the sync service, and the secure transport between devices. The interesting part is not making one lane work in isolation. It is making the boundaries clear enough that each lane can recover, stay observable, and keep moving without stepping on the others.

We are taking the time to validate the actual flows and keep the model honest as the product grows. More soon.