looking at this just makes me happy.

really good perspective whether building a startup or anything else, you want a path with stepping stones: things that work! for example, for me, an intermediate stepping stone is making $1M with an image&video generator

this is definitely a nod to Inception (Dreamina)

I can't stop imagining a future where work as we know it ceases to exist and instead, everyone works with super-powered AI agents that can do anything Suddenly, everyone has their personal team You're only limited by what you tell the agents to do (and perhaps even that won't

I remember back in the day, when Anthropic first introduced their Constitutional AI, I thought it was BS But now, after learning more about it, it seems that they were right after all. it seems that, after a certain point, the models can actually improve via self-evaluation &

Pretty funny idea Instead of optimizing the network, you optimize the *input* (or noise) that goes into the network Why? If you imagine the network as f(anything)->image: If you fine-tune the network, you risk going off the image manifold & producing some weird shit But if

TIL an image resizing app (generative fill) makes more money than Krea, Hedra and Higgsfield combined truly in the early innings

is it sad that the latest & greatest in aerospace is by private companies instead of governments? or inspirational

Random gens from the weekend. All Gen-4.

you know what, I'm buying the App Mafia full membership Zach Yadegari is a legend and I bet in 10 years, everyone will have wished they could buy this membership for only $400/mo To all the "these guys are grifters" folks, let me tell you why it's an insanely good deal:

Jared Hardin Moses Kagan Just think of mentors you had in the military. Approach with kindness and the sky is the limit. People love to help others.

AI video just keeps getting better! Today's launches (lipsync) - Sync is great for real-life edits - Wan 2.2 is about to drop a lipsync model (pure generation) characters: - nano-banana for consistent characters - wan/veo for animation Can't wait to make this easy to use

i'm so excited about gemini this is what happens when you have all the top researchers and make them NOT PUBLISH

pretty soon they're gonna be generating composite gaussians & the worlds will have movable/changeable assets rendering is there, diffusion is there, only a matter of time

these lofi mixes are getting crazier youtube.com/watch?v=pmnv2J…