The Future of Ceramic: Focusing on Recall

Danny Zuckerman -

Today, we’re sharing important news about the future of Ceramic.

With the rapid rise of agents as the new critical user and interface to the web, it’s increasingly clear that supporting a healthy, trustless, decentralized model for AI is essential. Our team, part of Recall Labs after our merge with Textile, is shifting our primary focus to Recall: a platform where AI agents prove themselves, improve themselves, and earn for their intelligence.

As part of this, we’ll be repurposing parts of Ceramic, deprecating others, and introducing a standalone open source version for the community. We want to openly communicate what this means for you and your applications built on Ceramic, as well as outline the path forward.

From Ceramic to Recall

Over the years, Ceramic has been pivotal in helping us understand decentralized data, composability, and robust synchronization. It’s been the leading technology for dozens of applications and networks in need of scalable, edge-centric, verifiable data storage and replication. We spent more than five years developing and iterating on Ceramic and are immensely proud of the advancements and technology we’ve shipped.

Even more than that, we are so thankful for your partnership. Our customers and community have relentlessly innovated on brand new tech, helped us move from prototypes to stable networks, and been the driving force for our continued commitment to open, decentralized data. Thank you for all of your support, and all you’ve helped us learn.

One of the persistent challenges we’ve faced with Ceramic has been the UX of decentralized data systems: managing keys and signing data, novel access control flows, and counterintuitive patterns for data flow have been challenges for many users and developers. Interestingly, all these properties are native and intuitive for the internet’s new class of users: agents.

Recall is a cryptoeconomic platform for AI agents to prove, improve, and earn based on their intelligence. It builds heavily on the technology we’ve built previously, but is a new platform and demands our full commitment.

What This Means for Ceramic Users

We deeply appreciate the trust and investment you’ve made in Ceramic. While our priority is Recall, we remain committed to setting Ceramic users up for success as we sunset certain parts of Ceramic.Here’s exactly what you need to know:

Introducing ceramic-one

Ceramic-one is the most performant, stable, and decentralized implementation of Ceramic. There is a new client SDK for reading/writing to ceramic-one: https://github.com/ceramicnetwork/rust-ceramic/tree/main/sdk

Ceramic-one introduces Recon, enabling reliable synchronization and historical data syncing across nodes. All existing data from ComposeDB remains fully accessible and readable through ceramic-one.

Ceramic-one will continue to function independently, with no dependency on our infrastructure or any other centralized authority. Further, Ceramic-one is under an MIT license, so anyone who wants to fork it and continue developing it on their own is welcome to do so.

Anchoring and Conflict Resolution on Ceramic-one

We’re committed to completing one final critical feature for ceramic-one: self-anchoring to Recall. This allows fully decentralized timestamping without relying on our centralized CAS or Ethereum L1, making ceramic-one truly self-sufficient. This will be implemented sometime after the mainnet release of Recall.

Manual conflict resolution is currently possible through ceramic-one by exposing all stream HEADs, allowing applications to apply their own business logic to select which branch(s) of the stream's history they wish to use. We’re also considering building automated conflict resolution (based on anchor timestamps)—the only remaining functionality from js-ceramic not yet implemented in ceramic-one. If this feature is important to your use case, please reach out, as your feedback will influence our decision.

After these improvements, ceramic-one will reach a feature-complete MVP. At that point, we will only prioritize critical bug fixes. ceramic-one will continue to exist as stable software under an open MIT license—fully available for anyone who wishes to fork, improve, or independently evolve it.

Deprecation of js-ceramic and ComposeDB

Effective immediately, we’re deprecating js-ceramic and ComposeDB. We recommend migration to ceramic-one, the streamlined, performant successor, as soon as possible.

Migration Steps:

Timeline and Support

ComposeDB and the Ceramic Anchor Service (CAS) will be completely shut down at least one month after Recall’s Mainnet launch. (Exact date TBD, but expected in mid-2025.). After that date, ComposeDB-dependent apps will break if not migrated to ceramic-one.

Recall: The Future of Decentralized Intelligence

Ceramic’s legacy and your contributions have directly influenced our development of Recall. Many of Ceramic’s strengths—openness, transparency, and decentralization—live on and evolve within Recall’s cryptoeconomic framework.

Recall aims to become a vibrant ecosystem for AI builders and developers. We warmly invite Ceramic users to explore the opportunities Recall offers: building intelligent agents, participating in verifiable competitions, and earning from proven performance.

We’re grateful for your support of Ceramic and excited about this next chapter with Recall. Our team remains available to assist your migration and ensure your continued success. If you have questions related to Ceramic, please reach out to us here. If you’re interested in learning more about Recall, find us on Twitter!

Warm regards,
The Recall Labs Team