👋 Product Hunt,
I am Gavin, the author of moar.
Virtually from the instant I began the usage of AI gear, I stored hitting the similar wall: I might drop a protracted file into Claude or ChatGPT and get “document too huge” or watch reaction high quality drop since the fashion was once wading via bloated inputs.
After I opened the supply information of those medical doctors, I discovered layers of formatting metadata, repeated headers, empty cells and different muddle the AI needed to learn however did not in reality want. I used to be paying for a top class subscription and burning maximum of my context window on natural noise.
I began performing some digging and got here to understand that our paperwork are formatted and designed for human verbosity, no longer AI comprehension. After which it hit me: to recover effects for a human like me, I had to first make it higher for AI.
So I finished guessing at what AI gear want from paperwork and I requested them. moar’s optimization algorithms had been co-designed with ChatGPT, Claude, Gemini and different LLMs. Each optimization step examined in opposition to what the fashions in reality use as opposed to what they forget about.
Thru a chain of structured conversations, side-by-side output checking out and iterative comments loops, they informed me precisely which formatting metadata is useless weight, which structural cues topic, the place bloat prices them consideration.
The result’s as much as 95% token financial savings with 0 lack of which means for AI gear.
Non-negotiables from day one:
– moar is 100% personal via design. No server, no uploads, no logs. Your paperwork by no means depart your instrument.
– moar’s AI-native file optimization engine is and at all times can be loose.
Would really like your comments, particularly bizarre information or edge instances I have never observed but. Truthful critique very welcome.
