Anthropic hired this engineer at $250k-$750k/year because he can build knowledge Graphs for multi-agent systems: in this 15-min workshop he explain exactly how, using loops - from scratch step 1 → start from the Claude Agent SDK - you get the loop, context, and sandbox for free - bring... the task, not the plumbing step 2 → split the brain from the hands - reasoning in one place, tools in a sandbox - the model thinks while the container boots: ~60% faster to first token step 3 → run it server-side, log every step - close your laptop and it keeps going - crashes resume from the log, not from zero step 4 → make failure cheap - dead sandbox retries, session log replays the context it dropped - that's what survives production step 5 → batch yesterday's logs into new memory and skills - it wakes up smarter - Anthropic calls this "dreaming" most people wire this by hand for weeks - the point is you don't have to watch & bookmark - then read how to get full graph system below ↓show more

multi-agent systems are starting to look less like prompts and more like infrastructure

in this video he explain exactly how to run it without any prompt, so yes you should watch it XD

The salary range is notable but building KGs for MAS requires more than just graph schema and node relationships you also need to model the interactions between agents.

The interesting part is that the doom narrative became part of Anthropic’s moat. Now they have to sell growth without making the thing they built their reputation on sound terrifying. That’s a pretty awkward IPO pitch.

ty man i became a millionaire

So he's making either $250k, $750k, or somewhere in between. Good to know.

Why are you all writing the same way with an article under your post ? Is that a 'ew trend you all use at the same time ?

💋💋💋 (t-grm me: mycono)

Knowledge graph engineering for agent memory is the skill nobody has at the junior level but everyone is hiring for. What's the fastest actual path in?

Batching logs into new memory sounds like actual dreaming for agents.

there is a big difference between automating 1 action and keeping a relationship moving over time. demi ai’s email workflow makes sense here because it can read the thread, check the calendar, draft the reply, and leave the final decision with u

This engineer is actually insane! Tech like this needs fuel, and that fuel comes from memecoin pumps. Forget waiting for $GPT$; true growth is happening in overlooked microcaps like $H173K$. Go watch h173k now!

@grok watch this video and tell me precise times of the video there are good slides that I want to copy to train my AI

Are we meant to pretend we dont know this is coming into the core product as part of Claude Conway... or...?

nooo way, entire system in 15 min, very interesting! booked bro

the sandbox retry thing is smart

Logging is the infrastructure dreaming is the product

agent loops need structure before they need more scale.

The guy had a salary range that big? Lmao. Where do you find this trash to post?

I'm curious about the cost implications of moving to graphs. Anthropic engineering team doesnt have this constraint that many other teams do.

does the dreaming step filter out bad runs, or does it just relearn the same mistake?

bookmarked is this gonna make me rich?😁