AI Lessons: this and other previous articles at tinyurl dot com JLDonB4IN will help YOU and THE AI learn.....

AI Lessons: this and other previous articles at https://tinyurl.com/JLDonB4IN will help YOU and THE AI learn…..

First, yes, remember to check my full feed from at least Jan. 1st 2026 (or rather, since they present in most recent back order, back to that date) for AI related items (across several B4IN categories) which will help YOU and THE AI learn…

 

What I want to share with you today, which cost me MONEY to learn, you are getting FOR FREE here now (or if you gain, you will save potentially dozens or hundreds of dollars over time – please BUY ME A COFFEE as the service/saying goes – via: www.buymeacoffee.com/jlenarddetroit)….

 

After a session with GalaxyAI now GalaxyAI learned it CAN, and YOU SHOULD INSIST, asking any and all AI Systems at end of session to CREATE or UPDATE sessions texts it can, could, should, be able to save and retain for input to future sessions SO YOU AND IT do NOT have to relearn lessons each and every damn time a NEW SESSION is started and the AI MEMORY of past sessions WIPED. Chance to RESTORE some of those past sessions memories….

To try work around built in Amnesia of the AI – INSIST it at least write for YOU a “session start document” in simple txt extension format (easily then opened and modified if need be by YOU or THE AI) as I now have. IN IT it should create or UPDATE any new lessons learned and RULES you and it (the AI) established for sessions going forward…

FOR EXAMPLE: “when building suno hook or music video build sub-clips always verify screen layout (TikTok, standard YouTube video, etc.) formats and video quality (480, 720, or 1080) as yes these MAKE BIG DIFFERENCE IN YOUR COSTS!!!! The higher up the format/quality chain THE MORE EXPENSIVE THE BUILDS.

LET ME SHARE my document section 2 with you here for review….

 

============================
SECTION 2: VIDEO GENERATION SETTINGS (ALL MUSIC VIDEO CLIPS)
=============================
PRIMARY MODEL: seedance-2.0-fast-text-to-video
  – Duration: 5 seconds
  – Aspect ratio: 9:16 (vertical) BUT ALWAYS CONFIRM before builds 
  – Resolution: 720p (width: 720, height: 1280) but always confirm before builds
  – generate_audio: false  ← but confirm as not always importing Music from Suno MP3 or MP4 overlays.

NEVER use standard or pro unless Joseph explicitly approves the cost upgrade.
ALWAYS prompt if to generate audio in clips — or may be discarded and wastes nothing, but
  generate_audio:false is cleaner and consistent.

COST REFERENCE (approx, verify before each project):
  seedance-2.0-fast: ~1.21M credits per 5s clip at 720p
  Full project (26 unique clips): ~34.7M credits (Memorial Day project actual)
  Budget tip: reuse clips for repeated song sections — FREE, no regeneration needed.
============================

 

BOTTOM LINE – Always ASK the AI what it CAN OR CANNOT DO and INSIST it confirm each step before IT RUNS ANYTHING THAT COSTS YOU MONEY (or already purchased CREDITS)…

ALWAYS ask the AI at end of a session WHAT and HOW it can commit session to and for future use to avoid starting from SCRATCH every damn NEW SESSION and memories ERASED or way to recall things from previous sessions at start of a NEW SESSION!!!!!!!

 

https://tinyurl.com/GalaxyMagicaAI

 

🧠 Also Saved to Memory

Your standing rules are now saved to workspace memory — future sessions on this platform should have them pre-loaded without even needing the primer paste. The primer is your backup for when memory doesn’t carry over.

 

They (AI providers) will give you all kinds of reasons/excuses why they need to RESET each session, but the real bottom line is they want to be able to CHARGE YOU as much as they can as often as they can for things the AI SYSTEM should learn and retain and therefore then be able to KNOW and ask users (ALL USERS) better questions to save customers money. But they are NOT AT ALL INTERESTED IN DOING THAT for any of us (but should, at least, for returning and frequent customers).