the insider knowledge nobody tells you
reaction control — how you send
→Don't send one-word replies like "yes" or "ok" — every message costs atoms. Batch your thoughts before hitting send.
→Skip "thanks" and "got it" messages entirely. Just send your next question.
→Don't ask Og AI to repeat or confirm what it just said. Trust it and move on.
→Describe the full change you want upfront. Iterating one tiny thing at a time burns atoms fast.
→Add "be brief" to your prompt. Cuts response length in half without losing quality.
→Bullet points in your prompt = tighter responses. Paragraphs invite longer answers.
→Ask for code OR explanation — not both — unless you genuinely need both.
mass & density — what you feed it
→A 30-page PDF burns tens of thousands of atoms just to read. Paste only the section you actually need.
→Images cost thousands of atoms each to process. Only upload when it genuinely changes the answer.
→Don't re-upload files you already shared earlier in the same conversation — they're still in context.
→Spreadsheets with hundreds of rows: share only the relevant rows, not the whole file.
→Summarize long documents yourself before uploading. A 3-sentence summary beats a 20-page paste.
half-life — when & how long
→5am–11am is peak server time. Slower responses, higher chance of hitting limits mid-session. Late night and weekends are fastest.
→Long conversations quietly degrade. Og AI starts compressing old context — key details get fuzzy. Start fresh when the topic changes.
→Put the most important info at the top AND bottom of long messages — never buried in the middle where it gets compressed.
→New topic = new reaction. You get a full atom budget back and a clean context.
→"Think step by step" costs more atoms but gives better results on hard problems. Skip it for simple ones.
isotope storage — memory, context & rules
The biggest atom waste: re-explaining the same preferences, project context, and rules every single conversation. There are four places to store things permanently — and most people don't know any of them exist.
The Formula — always on, every conversation
Your standing instructions. The things that never change. Write it once — Og AI applies it automatically every time, forever.
The Lab Book — lives with your project
A notebook inside your project folder. Og AI reads it automatically every time you open that project. No re-briefing, ever.
The Handoff — prime before you start
A quick note you hand Og AI at the start of a session to get it up to speed instantly. Solved something brilliantly last time? Reference it next time instead of re-solving from scratch.
Protocols — one word, whole workflow
Things you do over and over — deploy, review, generate, summarize — saved as a single trigger word. One word fires the whole thing.
Og AI notices when you've explained something three times and tells you — "this is worth saving. Want me to write it to a file?"
Og
AI
our approach
Bring your questions.
We take it from there.
no tiers of thinking
Every user gets the same intelligence. The same reasoning. The same quality of answer.
no hidden knowledge
The atom counter, the tips, the timing — visible from message one. Not buried in a help article. Not something you stumble on after six months.
no learning required
You shouldn't have to learn AI. Ask the question you actually have. Og AI handles the rest.
Everything they keep behind the curtain —
we hand to you on day one.
The atom counter. The tips. The timing. The things other users had to discover on their own — after months of burning through conversations and money the hard way. You get all of it the moment you arrive.
You were always supposed to know this.