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
- Tweets5.4K
- Following255
- Followers347K
- Likes5.9K
Pinned Tweet
Featured in @Forbes: Walrus Memory.
“There was a point in time that we were stopped from accessing Fable and Mythos. And imagine at that time, what do I have to do?” — @kostascrypto
5,000+ agents and 4.5M+ retrievals later, here is why builders are moving to portable, user-keyed memory so switching LLMs doesn't mean starting from zero.
Read the full piece by @guneyyildiz: forbes.com/sites/guneyyildiz…
ALT If a company can swap one model for another in an afternoon, the real lock-in moves to everything its agents have learned about the business. Chalkias noted three reasons companies switch models: a better model arrives, a cheaper one does the job, or a regulator or employer restricts which systems can be used. He pointed to this summer’s interruption, between June 12 and July 1, when access to Anthropic’s Fable and Mythos models was suspended. “There was a point in time that we were stopped from accessing Fable and Mythos. And imagine at that time, what do I have to do?” Chalkias said. In most enterprise setups, switching models means leaving accumulated context behind. “You’re not able to go and get your data and move it to another LLM.”
Walrus retweeted
The Total Recall Hackathon is still live through November 10.
Build across 3 tracks, wire Walrus Memory on Mainnet, and compete for $4,500 in prizes.
All builders welcome.
Details below in @tatum_io's post. 👇
D-1
Build from anywhere & compete for a share of the $4,500 prize pool 💰
tatum.io/hackathon/total-rec…
The Total Recall Hackathon brings together the @tatum_io AI Builder & @WalrusProtocol to help you build apps that remember.
Walrus retweeted
DECENTRALIZED STORAGE IS MORE IMPORTANT THAN YOU THINK.
Let's talk about @WalrusProtocol and why it's worth watching in the @SuiNetwork ecosystem.
Walrus is a decentralized storage and data availability protocol designed to help developers store and retrieve large files for Web3 applications.
Here's what makes it interesting
Built for data: Walrus supports storage for files, media, datasets, and data used by decentralized applications.
Powered by $WAL: The token is used for storage payments, network staking, and governance.
Liquid staking: Through @HaedalProtocol, users can stake WAL and receive haWAL, a liquid-staking token representing their staked WAL.
As AI agents and Web3 applications continue to evolve, reliable data infrastructure will remain an important part of the picture.
Walrus is building in that direction, and I'm keeping an eye on its ecosystem.
Do you think decentralized storage will become one of the biggest narratives in Web3?
#Walrus #WAL #Sui #DePIN
@OpenAI retired legacy API models including GPT‑4, GPT‑4 Turbo, o1, o3-mini, and o4-mini on October 23.
GPT‑4o and GPT‑4.1 willfollow on January 14, 2027.
Model lifecycles are normal. Your agent’s memory shouldn’t share them.
When memory lives outside the model layer, teams can change models without rebuilding accumulated context or locking it to one provider.
Switch models. Keep the memory.
With Walrus Memory, the interface stays the same: remember() on write, recall() on read – whichever model or runtime comes next.
developers.openai.com/api/do…
Thank you to all the builders who participated and shipped applications with persistent memory.
Here are the winners across the App, Article, Infrastructure, and Ecosystem categories. 🧵👇
Three ways to integrate Walrus Memory:
1. Core SDK
Store and recall memory with automatic embedding, SEAL encryption, and Walrus storage.
pnpm add @mysten-incubation/memwal
2. OpenClaw plugin
Add automatic memory capture and recall to OpenClaw agents, keeping memory available across sessions.
openclaw plugins install @mysten-incubation/oc-memwal
3. Vercel AI SDK
Add persistent memory to streamText and generateText workflows with middleware.
import { withMemWal } from "@mysten-incubation/memwal/ai";
Full setup + source on @github: github.com/MystenLabs/MemWal
Walrus retweeted
Featured in @Forbes: Walrus Memory.
“There was a point in time that we were stopped from accessing Fable and Mythos. And imagine at that time, what do I have to do?” — @kostascrypto
5,000+ agents and 4.5M+ retrievals later, here is why builders are moving to portable, user-keyed memory so switching LLMs doesn't mean starting from zero.
Read the full piece by @guneyyildiz: forbes.com/sites/guneyyildiz…
ALT If a company can swap one model for another in an afternoon, the real lock-in moves to everything its agents have learned about the business. Chalkias noted three reasons companies switch models: a better model arrives, a cheaper one does the job, or a regulator or employer restricts which systems can be used. He pointed to this summer’s interruption, between June 12 and July 1, when access to Anthropic’s Fable and Mythos models was suspended. “There was a point in time that we were stopped from accessing Fable and Mythos. And imagine at that time, what do I have to do?” Chalkias said. In most enterprise setups, switching models means leaving accumulated context behind. “You’re not able to go and get your data and move it to another LLM.”
Walrus retweeted
Bigger data is now part of agentic work on Talus
With Talus Protocol v2.1, files up to 8 MiB move through @WalrusProtocol, and every file can be traced back to where it came from, so tampered data gets rejected before an agent uses it
Here’s what else shipped in the release 👇
Your agent reads last quarter's spec and acts on it. A teammate would ignore it.
Andy Triedman of @Theoryvc joined @ConorBronsdon on why agents run short on context long before they run short on intelligence.
Walrus Memory gives them context they can carry and verify.
@Theoryvc partner Andy Triedman: the models are good enough. Most of what you'd ask an agent to do isn't intelligence limited, it's context limited --> stop asking for one integration, build the system that builds all of them
Andy also reminds us that an agent takes everything it reads as true, including the stale PRD a human would know to ignore - its our job to engineer systems that fix that.
Andy invests in AI infrastructure and applications at @Theoryvc and runs his own software factory. He joined me to discuss:
- building systems that solve whole classes of problems
- why a one-paragraph spec plus 3-4 follow-up questions is one level too high
- sending well-scoped tickets to cheaper models
- simulated customer environments for agents
- hierarchical agent memory, from session traces to company priorities
& more
Chapters:
(0:00) Agents are context limited, not intelligence limited
(0:34) From coding agents to software factories
(5:53) What still needs human approval
(10:16) Agent loops that improve the product on their own
(14:52) Where agent automation breaks down
(17:59) The bitter lesson and engineering as experimentation
(20:08) From writing code to directing agent fleets
(24:18) Shared context and stale decisions
(27:25) Inference ROI and cheaper models
(30:23) Moving human time from code to customer intent
(34:42) Sandboxes and simulated customer environments
(37:26) First steps: shared skills and cloud agents
(40:14) Protecting customer data and isolating agents
(42:32) Context limits and hierarchical agent memory
(47:05) What Theory Ventures looks for in founders
Brought to you by:
• @SvixHQ - reliable webhooks for startups and the Fortune 500. Qualified startups get $12,000 in credits, YC companies $50,000: link.svix.com/cot
• @WalrusProtocol - Walrus Memory gives AI agents portable, verifiable memory that carries context across apps, sessions, and other agents: walrus.xyz/cot
• @g2i_ai - a decade of vetting engineers, now turned on reviewing the RL environments, evals, and training data models learn from: fandf.co/3SFxVm6
• @inngest - durable execution for agents in production. Failed steps retry; completed steps are saved and skipped: inngest.link/cot-pod
Look up the @chain_ofthought podcast on YouTube, Spotify, Apple Podcasts (links below), or wherever you listen. Thanks, Andy!
Walrus retweeted
@Theoryvc partner Andy Triedman: the models are good enough. Most of what you'd ask an agent to do isn't intelligence limited, it's context limited --> stop asking for one integration, build the system that builds all of them
Andy also reminds us that an agent takes everything it reads as true, including the stale PRD a human would know to ignore - its our job to engineer systems that fix that.
Andy invests in AI infrastructure and applications at @Theoryvc and runs his own software factory. He joined me to discuss:
- building systems that solve whole classes of problems
- why a one-paragraph spec plus 3-4 follow-up questions is one level too high
- sending well-scoped tickets to cheaper models
- simulated customer environments for agents
- hierarchical agent memory, from session traces to company priorities
& more
Chapters:
(0:00) Agents are context limited, not intelligence limited
(0:34) From coding agents to software factories
(5:53) What still needs human approval
(10:16) Agent loops that improve the product on their own
(14:52) Where agent automation breaks down
(17:59) The bitter lesson and engineering as experimentation
(20:08) From writing code to directing agent fleets
(24:18) Shared context and stale decisions
(27:25) Inference ROI and cheaper models
(30:23) Moving human time from code to customer intent
(34:42) Sandboxes and simulated customer environments
(37:26) First steps: shared skills and cloud agents
(40:14) Protecting customer data and isolating agents
(42:32) Context limits and hierarchical agent memory
(47:05) What Theory Ventures looks for in founders
Brought to you by:
• @SvixHQ - reliable webhooks for startups and the Fortune 500. Qualified startups get $12,000 in credits, YC companies $50,000: link.svix.com/cot
• @WalrusProtocol - Walrus Memory gives AI agents portable, verifiable memory that carries context across apps, sessions, and other agents: walrus.xyz/cot
• @g2i_ai - a decade of vetting engineers, now turned on reviewing the RL environments, evals, and training data models learn from: fandf.co/3SFxVm6
• @inngest - durable execution for agents in production. Failed steps retry; completed steps are saved and skipped: inngest.link/cot-pod
Look up the @chain_ofthought podcast on YouTube, Spotify, Apple Podcasts (links below), or wherever you listen. Thanks, Andy!
Walrus retweeted
AI context shouldn’t die when you switch tabs.
"Data that you create in one tool can be used by another." @Kenton breaks down how Walrus Memory solves portable agent context so your data stays yours across every AI tool you use.
Interview + video by: @AltCryptoGems
AI context shouldn’t die when you switch tabs.
"Data that you create in one tool can be used by another." @Kenton breaks down how Walrus Memory solves portable agent context so your data stays yours across every AI tool you use.
Interview + video by: @AltCryptoGems
Walrus retweeted
Your AI’s memory deserves privacy
Lighthouse Memory supports encrypted context through Walrus Memory on @WalrusProtocol
Keep what matters
Pick up where you left off
Walrus retweeted
At Gotchi Labs, we see this as the foundation for the next phase of Consumer AI: Intelligence → Action → Commerce.
Let's talk, .@SuiNetwork and @WalrusProtocol teams! We'd love to explore how we can bring these capabilities into everyday Consumer AI experiences.
AI agents are getting smarter. The systems around them need to keep up.
Payments they can use. Data they can trust. Boundaries they can’t cross.
Five takeaways from Sui Basecamp on the infrastructure the agentic economy needs 👉 sui.io/blog/ai-lessons-from-…
Walrus retweeted
DATA is taking a big step with Sui.
250M+ receipts already recorded on Trace, a $30M investment from @SuiNetwork Foundation, and now the move to bring the @datafdn Network natively to Sui.
That’s a lot happening at once.
I like that this goes beyond just putting AI and blockchain in the same sentence. Rights cleared training data, verifiable provenance, and storage through @WalrusProtocol give this integration a clear purpose.
Now it comes down to execution. How well will these pieces work together, and what will the ecosystem build on top of them?
$DATA on Sui is definitely worth watching.
#DATAonSuiQuest
DATA is coming to @SuiNetwork.
The DATA Network and Trace are joining the Sui stack, bringing a live supply of rights-cleared AI training data to infrastructure built for AI workloads.
With 250M+ receipts already recorded by Trace, AI labs will be able to source real-world data with verifiable provenance, while @WalrusProtocol provides the underlying storage layer.
Sui Foundation is investing $30 million in the DATA Network, establishing the largest institutional stake in the network, alongside the continuing holdings of PIP Labs and the DATA Foundation. Furthering the alignment, SY Lee, founder of the DATA Network and CEO of PIP Labs, joins Sui Foundation as Strategic Advisor.
In the coming months:
▸ The DATA Network will be upgraded to run as a protocol natively on Sui
▸ Walrus becomes the default storage layer for DATA Foundation apps
▸ $DATA remains the token of the DATA protocol
The next chapter of DATA starts here.
Walrus retweeted
FIRSTIME WINNING BUG BOUNTY
THANKYOU SO MUCH @WalrusProtocol
AND SHIFU @cryptocymol
LETS FUCKING GOOOOO
The Total Recall Hackathon is still live through November 10.
Build across 3 tracks, wire Walrus Memory on Mainnet, and compete for $4,500 in prizes.
All builders welcome.
Details below in @tatum_io's post. 👇
D-1
Build from anywhere & compete for a share of the $4,500 prize pool 💰
tatum.io/hackathon/total-rec…
The Total Recall Hackathon brings together the @tatum_io AI Builder & @WalrusProtocol to help you build apps that remember.
Walrus retweeted
Perceptron x Walrus
Two infrastructure layers, one standard: enterprise data should come with proof of where it came from.
@WalrusProtocol is the data platform built for the demands of AI. Every dataset on Walrus gets a verifiable identifier and immutable history, held across a decentralized network that stays available even if up to a third of its nodes fail.
@PerceptronNTWK operates the acquisition layer on top: a network of 800,000+ nodes worldwide, turning live, ground-level data into structured signal for enterprises, funds and AI teams.
Perceptron is bringing its enterprise data suite onto Walrus. A decentralized acquisition layer needs a decentralized storage layer, and every buyer can verify nothing changed after collection. Walrus will also power our upcoming Data Questing Platform and data marketplace.
Live signal. Verifiable provenance.
Join the network: perceptrons.xyz
