Two new domains, two new sites, one honest read. What's working, and the handful of things I'd fix before these meet the world. Screenshots are current as of this morning.
"Watch the video" is the only call to action on the most gorgeous hero we've ever made. Anyone who doesn't feel like watching a video right now has nowhere to go. Add a quiet secondary button, "Meet the people" or "Explore the planet", so the curious have a path that isn't a play button.

The People grid is charming, but the cards fight each other: Mike gets a full scenic street, Chadwick gets a flat cream backdrop, Sally gets white, Dinesh gets a sunset city. Pick one treatment, either all scenic or all studio-flat, and the grid will feel like one family instead of six separate photo shoots.

Every visible post is from one account posting placeholder text. A first-time visitor reads this as nobody is here and nothing is happening. Before launch, seed the feed with real character voices: Sally on music, Chadwick on the week's outreach, Dinesh on what he's filming. Ten good posts beat fifty "tweet tweets".

0 replies, 0 reposts, 0 likes, on every post. Zeros don't read as "new", they read as dead. Either seed some honest engagement or hide the counts until a post has earned them. No number is better than a row of zeros.

The plan is that visitors click a character and chat with them. That's the product. But the homepage is a Twitter clone, the thing that makes LlamaChat different is nowhere on first load. Put a "Chat with" rail or hero strip of characters above the feed, so the magic is the first thing a visitor sees, not the fifth click.
"You're checking out Llama Chat, the main chat tool we use at the Studio. Enjoy :)", the winky emoticon undercuts an otherwise polished launch. Cut the ":)" and it reads like a product. Keep it and it reads like a draft.

Bottom line: E2 is a coat of polish from done. LlamaChat needs a content seed and its value prop moved front and center. Neither needs a redesign, both need a launch pass.