Pushing the frontier of memory and continual learning for agents. Add memory to your agents today
San Francisco, CA
Joined May 2024
- Tweets1.4K
- Following51
- Followers30.1K
- Likes1.7K
supermemory retweeted
supermemory does such gorgeous stuff
supermemory retweeted
we're glad we stuck with supermemory, since transitioning over, we never got a single complaint about Fae remembering stuff since! our signals for "forgetting" on @raindrop_ai dropped by 30% to single digits!
supermemory retweeted
Introducing supermemory's graph engine
yes. I did a graph. and yes, im against graphs. guess what
It's built ground-up from first principles (its not a real graph)
ALL you need for memory is:
- Keeping track of changing knowledge
- Forgetfulness on specific information
- Connections between information
We boiled the entire logic down into a single data structure. it can be represented as a filesystem, facts system, pages, or anything your agent / setup needs
This also allows us to magically find and inject tokens into the model harness without the model having to come looking for that information. Because
memory should be magical
and super intelligence needs @supermemory
supermemory retweeted
this is all it takes to get a @Lovable iMessage agent in production with @supermemory @browserbase @daytonaio BYOK
we put lovable in iMessage.
text an idea. build the app. ship it. send another message when you want to change something.
you can now build and launch software from the same place you text your friends.
introducing @onkeiki for lovable 🐎 💗
supermemory retweeted
one command and persistent memory for your coding agents 🧠
𝚗𝚙𝚡 𝚜𝚞𝚙𝚎𝚛𝚖𝚎𝚖𝚘𝚛𝚢 𝚙𝚕𝚞𝚐𝚒𝚗𝚜
supermemory retweeted
i ran some numbers. Markdown memory is 40x more expensive than @supermemory
let that sink in.
teams never realize this because token spend is not well attributed, so it feels like the users are using the model too much, when in reality it was just your memory consolidation step
supermemory retweeted
the @supermemory console walkthrough is up..
set it up, search it, connect it to your agents
i have a great respect for every youtuber, making long form videos are hard!
go watch it right now🫡
supermemory retweeted
Started using this with my fleet and think it’ll be a core part of my stack. The idea is to not be locked into any one harness.
T3 code for portable sessions,
@executorsh for portable MCPs,
@supermemory for portable memory, and Energon for portable artifacts.
It shouldn’t be hard to share docs and files between your agents. it shouldn’t be hard for your agents to create a doc for you to share with another human.
You shouldn’t need different services for markdown vs HTML vs prototypes vs zips.
It also shouldn’t be hard for it to not be locked into one vendor or harness.
A low key project a launched a few weeks ago is meant to solve this. I use every day with my agents. Each agent has the skill installed: Grok @bot, @Muse Instinct, Claude Code, ChatGPT Dot. You get the point.
Simple self hosted instructions in your own @Cloudflare account, works on their free tier too.
getenergon.com
supermemory retweeted
PSA to all founders. my @supermemory strategy:
drop open source gems, research, articles, etc. every day
build a great product, make lots of money
you can just do both™ if you got the dog
Introducing MemoryRepo.dev.
an open source implementation of Cognition's dreaming memory system, same as Instinct, Muse, Openclaw, etc., built on @cloudflare Artifacts, Durable objects, Alchemy and Effect.
I've been exploring agent memory for a long time and this is one of the more popular ways to do it, and I'm happy to see cognition putting out a detailed spec of how they perceive this.
In a way, it's beautifully simple and encapsulates what a good memory system should have.
There's many pros and cons, here's a video of me talking about it ↓
If you're thinking about agent memory you should probably use @supermemory - we think a lot about the very specific details about all of this. But, if you're curious, github.com/supermemoryai/mem… is a great place to start with the cognition memory thing as well.
we're actually going to incorporate some of the learnings from the research we're doing here into supermemory (soon-ish). but open sourcing it for public benefit :)
sorry for the bad audio quality, did it as a quick thing
supermemory retweeted
Small teams can take on engineering work they’d normally need a specialist for.
“It would have taken at least two to three weeks of active work.”
@supermemory's memory-graph optimization took a two-day Vorflux session and $600 in compute.
supermemory retweeted
We completely re-architected the supermemory API surface for agents
building the best (memory) api that agents have ever seen
supermemory retweeted
480k+ views and 650+ github stars in under 2 days!
we shut down company brain, to go all in on building the best memory system in the world, then open-sourced the whole multiplayer slack harness.
under the hood: one durable object per org on cloudflare agents, a haiku triage model deciding answer / investigate / react / pass, approvals that suspend and resume the same turn, and memory scoped per org, channel and dm on @supermemory.
quick look on some capabilities
supermemory retweeted
markdown vs @supermemory
in markdown only memory systems, the agent may have to look through and traverse files to answer things. This is wasteful and slow! The files also get stale really fast, and there's no forgetfulness, updates, etc. The agent has to just look through things.
supermemory is kinda like a dense interconnected markdown system, except that we automatically construct almost an 'instant' file that will directly one shot what the agent needs. That context has the whole history of the thing, removes forgotten info, and even related things that the model should probably know.
This information is dense, fast, and dynamically constructed, so you have a living breathing improving system.
So yes, you should use supermemory instead of markdown
supermemory retweeted
every day I'm even more bullish on how beautifully simple and composable supermemory's architecture is. And this is with incremental improvements throughout years of being with customers and really really caring about THEIR products.
@supermemory will be the default way to do memory in the agents era. There's nothing else that makes sense to me.
supermemory retweeted
Introducing the supermemory plugin for Muse code!
Introducing the supermemory plugin for @AIatMeta muse code.
Shared (super) memory capabilities that feels like magic!
supermemory retweeted
Big shout out to @supermemory for supporting memories so that the conversational interactions can feel much more personal for returning users.
Without their indexing this wouldn’t be possible.