Gemini 4 Argon can turn messy data into clean, visual narratives in just minutes 🪐 Here’s how it processed 80,000 files from NASA's open archives (CSVs, images, PDFs, web articles, XML logs, and raw videos) to build a Google Slides deck based on strict design guidelines.

Learn more ↓

@NASA 80,000 files and a Google Slides deck in minutes? My Downloads folder could never

@NASA The real benchmark isn't how good the slides look. It's whether the AI can extract the right insights from 80,000 files without missing critical details

@NASA Cool

@NASA Turning 80,000 raw files into a structured narrative is a serious test of AI's ability to synthesize information, not just generate text

@NASA Damm that's powerful. How long did it take to process those files?

@NASA This is where AI gets exciting: turning massive, messy datasets into clear stories people can actually understand. 🪐📊 Processing 80,000 files is impressive, but the real magic is turning that information into meaningful insights. 🚀

@NASA Wait, it handled random NASA file types and still whipped up a clean deck? I’m dying to know how it deals with weird file names or messy metadata. Bet there were some surprises in those archives.

@NASA Argon, 13 Ekim'den bu yana Gemini uygulamasında Google AI Pro ve Ultra abonelerine açık. Geliştiriciler için 1 milyon token'lık birleşik girdi-çıktı sınırı sunuyor; API'de 30 Kasım'a kadar milyon girdi token'ı başına 2 dolarlık tanıtım fiyatı geçerli.

@NASA Meanwhile my downloads folder is still the one archive no model has been brave enough to open

@NASA The fact that it ignored the 80k raw video files in your list is the only reason I’m not fully convinced. What did it actually pull from the footage?

@NASA The deck is the easy part to judge. With 80,000 files going in, I'd want every chart to link back to the exact CSV or PDF it came from.

@NASA did it read the raw videos or just the csvs

@NASA Impressive transformation of complex data into clear visuals.

@NASA Markets agree for now: Google is priced at 70% to still hold the #1 AI model at the end of October. The year-end race is much closer, 40% vs 45% for Anthropic.

@NASA 80,000 files into a clean deck in minutes is wild. the messy-data-to-story part is where most tools fall apart

@NASA Lol, I gave it four MD files and two excel files, it failed obey the guidelines and quota was gone as welll

@NASA 80,000 files and it still had to follow strict design guidelines. I've seen humans fail that with 8.

@NASA NASA explored the universe so Gemini could conclude, “Any questions?”

@NASA 80,000 files across completely different formats like raw video and XML logs is wild. Curious how much manual cleaning had to happen before feeding it in, or if it really handled the messy ingest directly.

@NASA turning massive unstructured datasets into polished presentation slides ignores underlying hallucinations and accuracy risks in critical domain applications

@NASA Turning massive, messy multi-format archives (CSVs, PDFs, raw videos) into strict design-compliant Google Slides is a game changer for workflows.

@NASA Honestly the hardest part here is following strict design guidelines. Most people I know can't keep one font through a 10 slide deck.