Loading video...

Video Failed to Load

Go Home

Agent tree component, see who started whom

226,460 views • 18 days ago •via X (Twitter)

27 Comments

Arek's profile picture
Arek18 days ago

Florian you’re unstoppable! 🔥 it looks like you’re creating the most interesting component library i’ve seen so far 💪

Flo's profile picture
Flo18 days ago

Really appreciate it Arek! I'm trying to adapt to the agents era 👀

Arek's profile picture
Arek18 days ago

and you're doing it very well, man! i take my hat off to you 🔥

Flo's profile picture
Flo18 days ago

Thanks a lot!! I really appreciate to hear that from you!

Stheno's profile picture
Stheno18 days ago

Would this make sense to do in t3 code? @theo

Sasha ༊·˚'s profile picture
Sasha ༊·˚18 days ago

this is insane!!! ✨✨

Flo's profile picture
Flo18 days ago

Thank you Sasha!!

kamran Ahammed Aman's profile picture
kamran Ahammed Aman18 days ago

How i can create

heyflo's profile picture
heyflo18 days ago

You're on fire man!

Razor's profile picture
Razor18 days ago

很酷

Nicholas's profile picture
Nicholas18 days ago

Mate, your components look great.

don0x0's profile picture
don0x018 days ago

Hold up. Is this just a UI component or a real product? I want an an agent orchestrator like this.

Atakan's profile picture
Atakan18 days ago

love this. we just shipped the memory side of the same idea: every memory shows who wrote it. who-spawned-whom + who-wrote-what is most of the debugging story for multi-agent setups

Limited Edition Jonathan's profile picture
Limited Edition Jonathan18 days ago

Ooh, I like this view. I have something similar going on, but in a way that I could actually respond to the agents, Answer questions and send new work. this is legitimately incredible.

Michel David's profile picture
Michel David18 days ago

nice!

Guilherme Lippert's profile picture
Guilherme Lippert18 days ago

I really prefer the T3’s Orchestrator v2 approach, not focusing too much on the subagents output BUT letting you see if you want Do we really need to know who started which agent or we should focus more on the quality of the output? Feels like babysitting to me, tbh, but I could be very wrong

Shawn's profile picture
Shawn18 days ago

This is sick man! I want to know if it works or will be working on mobile ?

Justin Bennett's profile picture
Justin Bennett18 days ago

Really cool, but I feel like it takes too much visual space. What if you used a compact dag visualization along the left and attached the agent cards to that?

Alex · AI & UX's profile picture
Alex · AI & UX18 days ago

the tree is nice for who started whom. i'd be curious how it holds up at 30 agents, do you collapse finished branches or fade them out?

3lite's profile picture
3lite18 days ago

looks so clean :)

ziziran's profile picture
ziziran18 days ago

For the stopped Tester, a link to its replacement (when there is one) would help distinguish a retry from an independent sibling task.

murloc.eth's profile picture
murloc.eth18 days ago

This rocks hopefully there’s a source 👀

Ravi's profile picture
Ravi18 days ago

love that the cost sits right on each card. once agents start spawning other agents, knowing which branch is spending the money is exactly what you want at a glance

Aapakari's profile picture
Aapakari18 days ago

Does it capture why each agent was spawned, or only who?

Paul Martin's profile picture
Paul Martin18 days ago

tasty!

eikkenberg's profile picture
eikkenberg18 days ago

No way you got a similar UX as me!! Yours feels much more polished though

Kashif Ali Khan's profile picture
Kashif Ali Khan18 days ago

visual tracing for spawned sub-agents is so clean man

Related Videos