Memory that
persists.
Raven is a decentralized memory layer on Walrus. Store context, train models, and give your LLMs persistent, verifiable memory — without centralized infrastructure.
Power to the builder
Use CasesContext that travels with your agent.
Autonomous agents lose state between sessions. Raven gives them durable, queryable memory — so every interaction builds on the last.
Learn moreTrain on verified data. Ship reliable models.
Store training datasets on Walrus with cryptographic proofs. Prevent poisoned data from corrupting your models. Trust your inputs.
Learn moreYour knowledge base, decentralized.
Embed and retrieve context from a censorship-resistant store. No single point of failure. Your data stays available, always.
Learn moreThree lines of code.
Connect your LLM
Integrate Raven's SDK into any model pipeline. Works with OpenAI, Anthropic, open-source — any LLM that takes context.
Store memory on Walrus
Context, embeddings, and training data are stored on Walrus's decentralized blob store — verifiable, redundant, always on.
Recall and evolve
Your model retrieves past context on every inference. Memory compounds. Your LLM gets smarter — session after session.
import { Raven } from "@raven/sdk"const memory = new Raven({ network: "walrus" })// Store context after each conversationawait memory.store(sessionId, context)// Retrieve on next inferenceconst ctx = await memory.recall(userId)llm.generate({ context: ctx, prompt })
Powered by Walrus.
Walrus is a decentralized storage network built for massive data blobs. Raven leverages it to give LLMs the persistence layer they've always needed.
Built on the Walrus Stack
Created by the team building on Walrus and Sui. Raven's modular architecture brings persistence, verifiability, and composability together so builders can focus on creating great AI — not managing infrastructure.
Simple, transparent pricing
Start free and scale as you grow. No hidden fees, no surprises.
- 1,000 API calls / month
- 1 tenant
- 50 users
- 7 days retention
- 25,000 API calls / month
- 3 tenants
- 500 users per tenant
- 30 days retention
- 150,000 API calls / month
- 10 tenants
- 5,000 users per tenant
- 90 days retention
- Analytics dashboard