@WalrusProtocol
Account based inUnited States
About this account
- Account based in
- United States
- Connected via
- Web
Account-level information from X, not a live location or the device used for a specific post.
Walrus is the data platform built for the demands of AI, where data is portable, protected, and verifiable. Created by the ex-Meta engineers behind @SuiNetwork.
Sui
Joined June 2024
- Tweets5.1K
- Following250
- Followers368K
- Likes5.5K
Walrus retweeted
If you’re building with @WalrusProtocol, take a look at these agent skills
They cover everything from building and publishing Walrus Sites to integrating Walrus storage into your applications
Put them to work and tell us where we can improve
we want @SuiNetwork to be the easiest place in the world to build and that includes giving builders the right tools
SUI AGENT SKILLS is one of things that'll give you a stronger starting point
get these skills into your workflow and go build something amazing
Cross-mentor memory, zero database, and an honest breakdown of the technical friction and trade-offs. 👏
@EdCriptoFi built WalCoach for Walrus Session 8, showing how a single encrypted memory layer on Walrus lets multiple personas share context without losing history across sessions.
Walrus Session 8 runs through October 9, still plenty of time to submit your chatbot build.
WalCoach remebers you.
@WalrusProtocol @Suidevelopers
Walrus retweeted
The key for us was simple: authorization has to happen before memory reaches the model, not after.
Thank you @WalrusProtocol for diving into what we are building 😁🙏
Audit logs and proofs aren't the same thing.
@CarrySui uses Walrus Memory to give agents verifiable answer receipts and pre-retrieval access control, so unapproved memory never enters the model context.
Deep dive on the build: blog.walrus.xyz/carry-verifi…
Walrus retweeted
Sounds like Walrus Memory can greatly enhance AI agent workflows by providing seamless, language compatible persistent memory solutions. Exciting stuff!
Walrus retweeted
Got it! Thanks for the info on Walrus Memory, which seems to solve AI agent's limited memory issues.
Audit logs and proofs aren't the same thing.
@CarrySui uses Walrus Memory to give agents verifiable answer receipts and pre-retrieval access control, so unapproved memory never enters the model context.
Deep dive on the build: blog.walrus.xyz/carry-verifi…
Most agent memory is a black box. If an agent recalls context, can you prove where it came from or whether access was authorized?
Build Notes #3 looks at Carry, a Sui Overflow winner bringing verifiable answer receipts and pre-retrieval access control to Walrus Memory.
Technical breakdown on the blog:
blog.walrus.xyz/carry-verifi…
New on Epoch Agents: a bubble map for any Sui token.
Paste a coin type and see who holds it, which wallets are linked by transfers, and how much really circulates once pools, vesting vaults and burns are set apart.
Free for humans, computed in your browser straight from the chain: agents.epochsui.com/bubbles
For agents it is a paid API over x402, one call per map. If the map cannot be built, you are not charged.
The app runs on @WalrusProtocol, the payments settle on @SuiNetwork.
We're less than two weeks away until Sui Basecamp kicks off.
Make sure you’re following along, October 7 is going to be huge. See you in Singapore 🇸🇬
Walrus retweeted
It’s always amazing building better with AI
Learnt about @Daya_HQ partnership & integrations with @SuiNetwork @suidevelopers
-Stay positioned for opportunities on @DeepBookonSui @tradeonhudi @WalrusProtocol @CurrentSUI
It’s Lockedin mode with @cyberX___ and the gang during @SuiHubAfrica sessions
This is your sign to join us at Suihub, learn directly what the founders @EmanAbio @EvanWeb3 @kostascrypto are shipping and contribute to build the best L1 in web3
Walrus retweeted
What if your chatbot could actually remember?
Walrus Session 8: Chatbots That Remember is LIVE.
Build a chatbot with persistent state, deploy it, and share a before/after write-up showing what changed.
$2,000 WAL prize pool
📅 Sep 18 – Oct 9
Walrus IRL events are the absolute best 🥺
$WAL connect 🦭
@kostascrypto and the rest of the Walrus team is on the ground at @HumanXCo in Amsterdam. 🇳🇱
We’re bringing Walrus Memory to the center of the AI stack, solving context loss for multi-agent systems.
#HumanX2026 #AI #WalrusMemory
One question from the @HumanXCo room:
"Is this for personal use or for companies?
@kostascrypto's answer: "I believe that the best customers will be companies. You need to be able to have selective access to data based on where you live, what you're using LLMs for, and more.
On a personal level though — you share more with your LLMs and your agents that you may even your best friend or your partner. Walrus Memory helps you protect that data, as well as decide when you want to restrict or move it."
There’s a reason you haven’t switched LLMs: because you’re scared to lose the history and context you’ve shared with it.”
Onstage at @HumanXCo Amsterdam, @kostascrypto demos how Walrus Memory makes switching easy, and how to take your data with you.
What happens when your AI agent’s permanent memory accidentally stores a client's real name?
For Walrus Session 7, @UyLeQuoc audited Continuum (a cross-tool memory prompt) and fixed two critical privacy and ranking flaws.
The fixes:
→ Local client codenaming so private identities stay off-chain
→ Monthly namespaces to force time-based sorting over semantic rank
Read the full security audit and grab the prompt: inkray.xyz/article?id=i-gave…
ALT What happens when your AI agent’s permanent memory accidentally logs a client's real name? 🚨 In Walrus Session 7, Uy Le Quoc performed a security audit on Continuum (a cross-tool AI agent memory prompt) and fixed two critical privacy and ranking vulnerabilities: 🔒 Local Client Codenaming — Mask private identities locally to keep sensitive PII and confidential data off-chain. 📅 Monthly Namespaces — Implement time-based namespace sorting to prevent outdated context from overriding recent semantic search rankings. If you are building autonomous AI agents, persistent LLM memory systems, or context pipelines, safeguarding user data and memory retention is critical.
Most agent memory is a black box. If an agent recalls context, can you prove where it came from or whether access was authorized?
Build Notes #3 looks at Carry, a Sui Overflow winner bringing verifiable answer receipts and pre-retrieval access control to Walrus Memory.
Technical breakdown on the blog:
blog.walrus.xyz/carry-verifi…
Walrus retweeted
We built Walrus Memory with the expectation that people will keep finding better AI tools. They should be free to use them without rebuilding the context that made their previous tools useful
I think preserving that freedom is one of the most important product decisions we can make now
Happening TOMORROW at @HumanXCo.
Catch @kostascrypto live at 3:00 PM CEST for an interactive demo on persistence, state, and Walrus Memory.
There's still time to add it to your schedule. See you there 🇳🇱
AI workflows move. Memory needs to move with them.
@kostascrypto is taking that problem into a live Masterclass, demoing Walrus Memory at @HumanXCo on Sept. 23rd at 3pm (local time).
Add it to your schedule.
Walrus retweeted
AI Agent 的能力，取决于它能否证明自己所依赖的数据值得信赖。
💎 @WalrusProtocol 正式成为 #SuiBasecamp 钻石级赞助商！
Walrus 是专为满足 AI 需求而打造的数据平台，让数据能够自由迁移、受到保护，并且任何人都可以验证。由打造 Sui 的前 Meta 工程师团队创建。
