I built a hype-check for AI launches. What would make you NOT trust a verdict?
Every week an AI launch shows up claiming "10x faster" or "beats every benchmark". Some of it is real. A lot isn't. Finding out costs 20-30 minutes of digging through benchmark tables and dev threads.
So I built Reality Filter. You paste in a launch and get one verdict: worth trying now, watch later, too early, or mostly fog machine - plus a fog meter for how much of the claim is still unproven. Every verdict names its sources: real benchmark numbers, named reviewers, actual user reports. Never "reports suggest".
Two things I did deliberately:
Verdicts are dated, public, and I don't quietly edit them. One already flipped from "too early" to "worth trying with caveats" when GA landed - and the original call is still sitting on the record next to the correction.
One verdict is "mostly fog machine" on a product doing $1.2B ARR. If this only ever agreed with press releases it would be worthless to you.
Free, no signup needed to read a verdict: https://bharatlearner18-del.github.io/reality-filter/
There's also a 22-second explainer I just added, because early visitors told me they opened the site and couldn't tell what it did: https://bharatlearner18-del.github.io/reality-filter/about.html
What I actually want from this thread, more than clicks:
When a launch you're unsure about drops, what do you do right now? Ignore it, ask someone, dig it yourself?
- What would make you NOT trust a verdict like this? I'd rather hear the reason it fails than a compliment.
Replies