Itās a bug on our side, we will fix it
Thanks for letting me know, glad I could help with finding a bug ![]()
I understand not being able to switch contexts in a given chat. What Iām asking is if we can start a new chat to modify that app, rather than when we click edit being forced into our original chat?
Sounds like an interesting idea, we will explore it
They are effective immediately. Itās just you get the entire month worth of credits to play with for the first ~week.
July will be the first proper month.
As you can find out from our blog post, we had optimized Fibery AI prior to the transition to credits as much as we could. I hope, weāll find more ways in the future, but for now the credits represent the real cost of unsubsidized calls to model providersā APIs.
With enough demand, weāll allow to pick the model for a given prompt. That way youāll have more control over cost vs. performance balance.
Great work to still release while you had the outage ![]()
Custom Apps - love it! And nice work with the permissions which make it easy.
I tested it by porting over an existing custom HTML page and it worked great. It was (almost) perfect in one-shot and only 1 one follow up prompt.
But it will only be really useful if we can build outside Fibery. I already have ways of working, tools, skills, versioning, etc and comfortable using apps like Cursor and Claude Code. And from a cost perspective LLM subscriptions are massively subsidised up to 40x cheaper compared to buying tokens.
AI Credits - great move and will be especially good when added to automation steps. Are there plans for wider range of models and/or BYOK?
Fibery AI permissions - hopefully this is one of first steps towards āAdmins can use it to check what access other users haveā? That would really help to set up permissions correctly and do audits. I know there is some UI but now everything that takes multiple clicks though unfamiliar menus I have to interpret seems so slow and difficult compared to just asking AI a very direct question.
Will you please let us know when the apps will be editable and available to the public? Thanks!
In the next ~3 weeks we will implement Apps versioning, enable export/import apps via Templates and open them for external AI Tools. Only after this we will remove experimental flag and enable it in all accounts by default. So I guess somewhere in the end of July it will be public.
Yes, we had many discussions around AI in automations, but postponed this area for 1-2 months, we will get back to it after Custom Apps and MCP improvements, right now we want to focus on these two themes
Yes, all this is in plans for the near future
Thanks for replies, really looking forward to this!
āIn the next ~3 weeks we will implement Apps versioning, enable export/import apps via Templates and open them for external AI Tools. Only after this we will remove experimental flag and enable it in all accounts by default. So I guess somewhere in the end of July it will be public.ā
Yay! Really looking forward to this!
I think it would be a smoother experience for Fibery to host a native importer/migration tool.
Iām going to piggy back on this request. I was hoping to land a time tracker as my first demo today (was not fortunate enough to have extra capacity in the last week of Q2) and burned through 500 credits rapidly.
I think itās prudent to allow us to burn through tokens on our external LLM subscriptions if weāre all out of Fibery tokens especially given that we donāt yet know how best to optimize requests/builds.