So I’m using Figma Make to develop an app, have been working on it for months now and everything has been going really well so far. I was backing up to Github and then pushing to Vercel to deploy to my app, was great, everything working smoothly, I had testers working on it and surprised at how it was running so well.
That was until the 22nd April (Sydney time) Figma Make decided to move all my files from the /src/ folder to a brand now /src/app/ folder without any notice and my deployments started breaking. It took me a while, but comparing github commits and noticing there was a huge file list in one of them being moved, I finally figured out the root cause of the problem. In the meantime I had tried to get Figma Make to move everything back, but the root cause of the problem was that Figma Make, no matter what work was done on the entrypoints and other import paths, etc, was ALWAYS pointing to the new structure of /src/app/.
I considered starting from scratch and had my previous commit files to copy up, but when I created a new base app to copy everything into, it defaulted to /src/app/ - which to me signified a change in the base settings of Figma Make that this was how they were going to structure their apps moving forward.
If I was starting from scratch, this change in their defaults would have zero impact on anything, but since it changed partway through my existing app, it completely broke everything and I needed to use up about 75% of my credits quota to get it working again.
This happened right at the end of my credits cycle and luckily I had some left over, but the majority of it has used up 75% of my credits within the first few days of my credit period and it’s taken Figma Support (of who i contacted straight away once I figured out the problem, providing screenshots of transcripts and vercel folder logs of before and after as well as gave them view access to my file a week ago) nearly 4 weeks to have their “engineers” come back and say that “the behavior appears to be a combination of both user-side workflow issues and normal LLM behavior, rather than a clear technical malfunction. From their perspective, the model actions themselves were not outside of what would currently be expected.”
So I have two questions…
Has this happened to anyone else? and;
Does anyone else know of the next steps I can take?
Sure, it’s only AU$30, but the fault is clearly on their side and I’m be damned if I’m letting them get away with this BS.