Every step of one project.
Written from a project we shipped: the steps that held, the traps, and the fix for each one. Your AI loads it and builds yours, and you make the calls.
One payment. It lands in your Vibedna library as one file, yours for life.
Drop the file into your AI chat, or connect Vibedna once and your AI reads it straight from your library.
Ask for what you want. Your AI follows it step by step on your project, and you make the calls.
What you can build with it.
Image, video and audio models running on GPUs rented only while a customer's job runs.
A hard spending cap, and a check that nothing is left running.
A job ledger, the cost of each job, and rules for where customer files live and for how long.
One job traced from click to link, and the steps to move a studio off a shared server.
16 chapters, in the order you build.
- 01When each makes sense
- 02Ground rules
- 03The hard spending cap
- 04Check that nothing is left running
- 05Worked example: one job from click to link
- 06The door between your site and the GPU
- 07Choosing where models run
- 08Cost per job, caps and plans
- 09Customer content: where it lives and for how long
- 10Moving a studio off one server
- 11When the GPU machine stays on
- 12Models, weights and storage
- 13Text and chat models
- 14Making it a subscription service
- 15What broke, why, and the fix
- 16Decisions and lessons
You bring your AI, your own accounts and your taste. The journal brings the steps, the traps and the fixes.
Comes with 2 start prompts.
Paste one into your AI and it knows exactly how to use this journal from the first message. They wait in your library, next to the journal.
Run Generative AI on Demand in the Cloud
One file your AI installs, and the whole method is in your chat.