The Context Cloud for agents is here. Introducing the new supermemory. A new way to do context, for your agents Everything you need for context - Memory, RAG, Filesystems, Profiles, are now available as blocks you can compose to build something for your use case

Try it today (we WILL personally help you improve your agents)

@supermemory yall got some good video production

@supermemory perks of having a @shardul_mane in the team

@supermemory @shardul_mane 🐐

@kushthaker03 @supermemory LFG!!

@kushthaker03 @supermemory yay!

@supermemory LEGO explainer team! Love it dude!

@supermemory NGL i lowkey copied the cloudflare stack video! a big inspiration

@supermemory Not copying, you just retrieved it from memory. My stuff is always opensource

@cneuralnetwork @supermemory this is huge

@supermemory Lego set for context engineering

@supermemory this is so fire 🔥 keep it up bro

@supermemory hell yeah

@supermemory LFG

@supermemory YES

@supermemory Lfg

@supermemory ABCDLFG (always be closing deals LETS FUCKING GO)

@supermemory 🫡

@supermemory looks promising. splitting those components into blocks could really speed up building customized workflows.

@supermemory support for cloudflare agents?

@supermemory already supported! Cloudflare agents builds on top of vercel AI sdk if im not wrong, which we have a native sdk for

@supermemory W video

@supermemory Can we plug in Pinecone as an external vector store for the RAG/Memory blocks?

@supermemory We have a vector store built into the cloud version! But on self hosted / enterprise versions you can swap out the vector store

@supermemory this looks sick, excited to test it out

@nexxeln @supermemory lfgggg

@supermemory 🔥🔥

@shivamhwp @supermemory time to make a switch, this is fire!

@supermemory And here I was thinking, supermemory was dead but ohhh bouy, what a comeback video. Hope so the product would live up to the hype you made in my mind

@supermemory Context is the real moat. I'm building @getalchemyst

@supermemory and i thought @karpathy second brain would work just fine for this, guess i was wrong

@supermemory I still remember bro's first video from the terrace starting a challenge to make money online. Great to see how quickly things changed

@supermemory i already built this and run this locally

@supermemory This feels like the right abstraction level. The interesting shift is from "memory as retrieval" to "context as runtime substrate" memory files profiles and workflows becoming composable blocks the agent can inhabit instead of tools it occasionally calls.

@shivamhwp @supermemory incredible ship

@supermemory This is an interesting shift where context is becoming infrastructure. A year ago most people focused on model quality. Now a growing part of the moat seems to be memory, retrieval, identity and orchestration.

@supermemory 云这种時候記得叫大家起床呢

@supermemory very nice video

@supermemory composable context blocks are the right instinct. the harder question is what happens when your memory block contradicts your filesystem block. each retrieves independently. you assembled the pipes. internal consistency across sources is a different problem

@supermemory Thought this was a farza video at first lol

@supermemory compose-the-stack shape is the right call. opinionated 'agent platforms' break the moment a use case has one weird requirement. blocks for memory + RAG + profiles let builders ship the slice they actually need, not the framework's worldview.

context is the real bottleneck for agents right now. most people optimize for smarter models when the actual problem is that the agent doesn't know enough about your codebase, your preferences, or what it did yesterday. the composable blocks approach makes sense because every use case needs a different mix of memory types.

@supermemory storing context was never my problem. eviction was. what gets dropped when the window fills, and who decides. does supermemory make that call or do i still have to babysit it

@supermemory 0xcdfa5bf5abd140e9d1c474c1b46d0fcc091edba3 this token going support you with fees cheers

@supermemory Supermemory to the moon!!