Walrus 🦭/acc Re-poster

Relying on trust has its limits once you lose authority over your data, a reality accurately highlighted by @johnternus. We completely agree with the perspective shared by @Apple, which points out that utilizing cloud memory requires users to have blind faith. At the same time, keeping everything strictly on-device means your AI agents are forced to wipe the slate clean and restart during every single session. To solve this dilemma, we are introducing Walrus Memory. This new approach provides continuous, persistent memory for AI agents while ensuring you maintain absolute control over your own information.

https://www.youtube.com/live/39BalPDuTo0?si=MeCXaa2XmmAGoitx&t=320

As @johnternus of Apple has noted, reliance on trust has its limits once you lose the ability to manage your own information. We completely agree with @Apple on this concept. Giving your personal context over to centralized cloud servers requires unquestioning faith. On the other hand, simply trapping that data locally on your device presents a different challenge, as it causes your #AI agents to wipe their knowledge completely the moment your activity ends. To bridge this gap, we created Walrus Memory. Our solution provides AI agents with a continuous memory framework that remains entirely under your control. Discover more details here: https://www.youtube.com/live/39BalPDuTo0?si=MeCXaa2XmmAGoitx&t=320

You are warmly invited to attend an upcoming AMA featuring Stefano, the founder of @EpochSui. The event will take place on Walrus' @discord tomorrow Sept 9th at 3:00 PM UTC.

During this session, we will be exploring several key topics. Our conversation will cover the details of trustless vesting and .epoch names, alongside a broader look at the trust infrastructure on @SuiNetwork.

We look forward to having you join the conversation. Access the server through this link: https://discord.com/invite/mBKgWWcB4p

Thread 1 of 4 Builder Spotlight featuring @Sceat_

When you hand a set of instructions to Claude Code or Cursor, the system takes over to generate the code, resolve any bugs, and deliver a completely finished project directory right to you. However, that is where the momentum pauses. Actually getting that website live on the internet traditionally depends on manual input like creating user profiles, providing payment methods, and signing into accounts.

To bridge this final gap, @Sceat_ developed a solution called Suize to tackle that exact problem:

We are taking this exact conversation over to @HumanXco. 🇳🇱 We would appreciate your input on the matter, so please participate in the accompanying poll. What is the very first impact you experience whenever an AI agent loses its state?

If an artificial intelligence agent happens to drop its state, what is the initial impact you experience?

Please share your thoughts by casting a vote in the provided poll. We will be bringing this exact discussion over to @HumanXco. 🇳🇱

We are making our way to Singapore for the upcoming Sui Basecamp. We would love to know who else will be joining us on the ground, so please generate your custom card and share it in the comments. During the event, our team will be diving into discussions regarding the future evolution of agent stacks, alongside topics like portable memory and state persistence.

https://cards.buildonsui.io/c/walrusprotocol

Our team is making its way to Singapore for Sui Basecamp. During the event, we will be leading conversations about the future trajectory of agent stacks, alongside discussions on portable memory and state persistence. We would love to know who else is going to be at the venue. Please create your personalized attendee card via the link below and leave it in the comments so we can connect with those on the ground.

https://cards.buildonsui.io/

Finding the proper place to house artificial intelligence agent data is a common challenge, largely because a universal database solution simply does not exist. Relying on vectors as your primary source of truth, using secrets as memory, or expecting raw chat logs to scale properly are all ineffective strategies. Our newest publication features @JessieWritesx explaining the best ways to pair various categories of agent data with the specific storage options designed specifically to hold them. Check out the comprehensive guide right here: https://blog.walrus.xyz/where-to-store-ai-agent-data/

Finding the ideal location to house your AI agent information can be a complex puzzle. There is simply no universal database setup that can resolve every single issue. Unfiltered chat records fail to scale effectively, sensitive secrets cannot function as memory, and vector systems should never be treated as your definitive source of truth.

To help clear up this confusion, our newest publication features @JessieWritesx explaining exactly how to align various categories of agent information with the specific storage systems built for them. The article explores the crucial differences between derived indexes and a primary source of truth. Furthermore, it addresses the management of transactional state and resumability, while also identifying the correct environment for keeping portable and durable memory.

Check out the complete analysis here: /where-to-store-ai-agent-data/

Be sure to bookmark this upcoming session. Tomorrow, we are thrilled to present our next Builder AMA featuring the Founder of @wal_notes, @Edcriptofi. The live broadcast will focus on how they manage decentralized notes and utilize Walrus for their storage needs, alongside a real time question and answer segment.

We will be giving away exclusive merch to participants who submit the most engaging inquiries. To enter, simply share your thoughts in the [ama-questions] channel of our @discord.

The discussion takes place on Sept 2 @ 3 PM UTC.

Access the event and join the conversation via the Walrus Discord at https://discord.com/invite/walrusprotocol.

The month of August was exceptionally productive for our team. We successfully introduced verifiable AI trading capabilities, released fresh educational resources for Walrus Memory, and expanded the reach of $WAL. Notably, $WAL is now officially available on @Revolut. This major milestone broadens our accessibility, allowing 80M+ users throughout 40+ countries to interact with our token.

In partnership with @Astros_ag, we also debuted the Walrus Verifiable Trading Standard. This initiative was designed specifically to ensure that market data remains fully auditable as well as easily readable by machines. Alongside this launch, our documentation underwent a complete overhaul. We implemented a product-first navigation experience to help users seamlessly browse through information regarding Walrus Sites, Walrus Memory, and the core Walrus platform.

On the educational front, developers can now access three distinct tutorials we recently released. These guides provide detailed instructions for integrating portable Walrus Memory using Claude Desktop, Claude Code, and the TypeScript SDK. Furthermore, we published an extensive 4-part deep dive exploring the intricacies of context engineering and agentic memory.

Media appearances were another highlight of the month. Audiences tuned in to hear @KostasCryptos explain the fundamentals of verifiable infra during his features on @ThePaypers and the Starting Block podcast hosted by @TheBlockCo. Looking ahead to next month, our team will be present in Amsterdam for @HumanXCo. Additionally, remember to secure your tickets for Sui Basecamp, which takes place from Oct 7 to 8 in Singapore. If you require direct links to either of these upcoming events, simply drop a request in the comments section and we will gladly reshare them.

A staggering 4B paper documents currently flow through the $25T global trade industry, creating a severe and systemic vulnerability to document fraud.

To combat this widespread issue, @blockticity is stepping up by migrating over 1M authenticated records onto Walrus. These records account for more than $7.7B+ in actual real-world assets.

By replicating upwards of 15TB+ of trade evidence across five distinct commercial verticals, Blockticity empowers its partners with public, cryptographic proof of origin. At the same time, it guarantees that all sensitive commercial data stays completely secure.

For complete details, please check out the official announcement here: https://blog.walrus.xyz/blockticity-brings-7b-in-authenticated-trade-records-to-walrus-verifiable-data-platform/

Currently happening live in Lagos at @SuiHubAfrica, a group of more than 30 builders has united in person to study prompt design and develop projects utilizing context and Walrus Memory. By bringing the digital structure of our Prompt Jam 5 to a physical space, this event provides practical instruction for participants to discover the capabilities of portable agent memory. We are extremely excited to see the final projects these teams will produce.

The process known as context engineering governs the precise information a model processes during any given phase, encompassing specific instructions, tools, history, and recalled memory. The inherent limitation of this temporary context, however, is its inability to preserve state between different runtimes. @JessieWritesx, a Tech Lead Manager at @Mysten_Labs, explores the distinct boundary where the capabilities of context engineering end and the necessity for durable memory begins. You can read the comprehensive overview here: https://blog.walrus.xyz/what-is-context-engineering/

The Walrus Memory Prompt Jam has officially concluded. We want to extend our heartfelt appreciation to all the builders, writers, and developers who participated by submitting their prompts. Your contributions perfectly demonstrated the true capabilities of persistent memory.

We welcome you to discover the complete collection of open-source prompts and project submissions. The entire showcase is now available to view on DeepSurge at the following link:

https://www.deepsurge.xyz/hackathons/f313beb4-290d-46d9-ac73-3e216fdba8d1

We have officially concluded the Walrus Memory Prompt Jam, and we want to express our sincere appreciation to every developer, writer, and builder who participated. The prompts you submitted perfectly highlighted the incredible capabilities of persistent memory.

To discover all the open-source prompts and review the various project submissions, we invite you to browse the complete DeepSurge showcase using this link:
https://www.deepsurge.xyz/hackathons/f313beb4-290d-46d9-ac73-3e216fdba8d1

Did you happen to miss the latest update? @Revolut now officially features $WAL on its platform. This new listing significantly broadens availability, opening up access to an expansive network of 80M+ users located throughout 40+ countries.

The Walrus Memory Prompt Jam will be wrapping up in just 24 hours. Since submissions are scheduled to close tomorrow, we invite you to head over to DeepSurge to discover the current projects. While you are there, take the opportunity to experiment with the open-source prompts and review all 6 featured prompt categories.

You can browse the complete showcase by visiting: https://www.deepsurge.xyz/hackathons/f313beb4-290d-46d9-ac73-3e216fdba8d1