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
Pinned Tweet
Introducing API v5, the agent-first memory API
- container tags are now namespaces, and they live in the url
- output from the API is now optimized for agent context
- one prompt to migrate
- Markdown negotiation in API results
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!
Introducing the supermemory memory graph
every document becomes atomic facts, and dreaming keeps building the graph in the background
- extracting facts
- linking related memories
- resolving updates, the old version stays as history
get one for yourself in the console ↓
read more here xitter.cf/DhravyaShah/status/211…
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
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 local v0.0.9 is out!
- API v5 support
- Better security
- about 40% faster
- you can use it with docker now
(we heard you! The next version of local will be 10x less resource intensive)
listen to @DhravyaShah talk about his cloudflare days, going viral on github, turning down early money, why we shut down products, ai companions and much more...
↓
Conversation with @DhravyaShah, Founder of SuperMemory
00:00 Cloudflare Culture & Working Under CTO
01:36 Autonomy, Writing Culture & Product Flywheel
02:45 Product Life Cycle: Idea to Launch
04:28 Demo Days & Engineering Writing Culture
06:32 Origins of Supermemory
08:00 Acquiring the Domain
08:59 Going Viral on GitHub & VC Interest
10:03 Turning Down Early Investment Deals
13:38 Early Investors & Susa Ventures
17:08 Shifting from Product to Infrastructure
18:49 The Market Street Epiphany & Shutting Down Products
22:04 Focusing on AI Memory & Continuous Learning
24:47 Inbound Go-to-Market Strategy
27:28 Testing Memory via AI Companions
28:11 AI Companion Dynamics & Psychosis Risks
30:31 Embracing Iteration Velocity
31:31 Lessons from Company Brain
33:25 Defining Good Memory Systems
34:32 Rethinking Memory Retrieval & Injection
36:45 Supermemory Architecture & Local Models
39:00 Memory for Continual Improvement
41:33 Agent-to-Agent Security Risks
43:32 Public Identity & Personal Growth
46:12 Productivity Philosophy
An UNFORGETTABLE #SFTechWeek, the Supermemory way!
We teamed up with @agi_inc to host a pitch night, a robotics hackathon, and a café pop-up.
Loved seeing what everyone is building, hearing the pitches, and getting to know people over coffee.
Thanks to everyone who came through!
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🫡
