LangChain Hyperspace Integration: Spatial AI Memory

Give your agents reflex-level speed and spatial reasoning.

This is the official LangChain integration for HyperspaceDB — the world's first spatial AI engine designed for Autonomous Agents, Robotics, and Continuous Learning.

🧠 Beyond Vector Search: Spatial AI Memory

Traditional vector databases are built to search static text. HyperspaceDB is built to model human cognition and physical world hierarchies:

• Fractal Knowledge Graphs: Euclidean vectors fail at hierarchies. Our native Hyperbolic engine (Poincaré & Lorentz models) compresses complex codebases or taxonomies into low-dimensional spaces, reducing RAM usage by 50x without losing semantic context.
• Continuous Reconsolidation: Transform raw information into episodic memory. Use built-in Flow Matching and Riemannian Math (Fréchet mean, parallel transport) natively within your LangChain chains.
• Edge-to-Cloud Sync: Robots and web-agents can't wait for cloud latency. Use the Merkle Tree Delta Sync protocol to handshakes episodic memory chunks between local devices and the cloud.
• Lock-Free ArcSwap Architecture: Built on Rust. Achieve up to 12,000 Search QPS and 60,000 Ingest QPS with near-zero latency, even under extreme agent concurrency.

📦 Installation

npm install langchain-hyperspace hyperspace-sdk-ts @langchain/core

🛠 Usage

Hyperbolic Memory Initialization (Poincaré Ball)

import { HyperspaceStore } from "langchain-hyperspace";
import { HyperspaceClient } from "hyperspace-sdk-ts";

const client = new HyperspaceClient("localhost:50051", "YOUR_API_KEY");

const vectorStore = new HyperspaceStore(
    embeddings, // Your favorite embeddings or useServerSideEmbedding
    {
        client,
        collectionName: "agent_spatial_memory",
        metric: "lorentz", // Use hyperbolic geometry for hierarchical knowledge
        dimension: 64,      // High semantic compression
    }
);

Advanced Spatial Pruning (Geometric Search)

Go beyond simple similarity. Prune memories using spatial regions:

const results = await vectorStore.similaritySearch("Find drone flight patterns", 5, {
    spatial_region: {
        $in_ball: {
            center: [0.12, -0.45, 0.88, ...],
            radius: 0.15
        }
    }
});

🎯 Enterprise Query Planner & Intent Disambiguator

Pre-evaluate user queries before issuing vector retrieval. Automatically splits compound queries into parallel searches and produces clarifying questions when queries are underspecified:

import { HyperspaceQueryPlanner } from "langchain-hyperspace";

const planner = new HyperspaceQueryPlanner();
const plan = await planner.call({
    raw_user_query: "Compare HIPAA vs GDPR retention limits and how do we encrypt client records?",
    collections: ["compliance_policies", "security_docs"]
});

if (plan.status === "clarification_needed") {
    console.log("Ask user:", plan.clarification?.question);
    console.log("Options:", plan.clarification?.options);
} else {
    console.log("Execute parallel searches:", plan.search_queries);
    // [
    //   "Compare HIPAA vs GDPR retention limits",
    //   "how do we encrypt client records"
    // ]
}

🕸 Native Vector-Graph (KAG) & Topology Linking

Connect LangChain document nodes and entities directly in the in-engine graph:

// Explicit directed linking
await vectorStore.link(1, 2, "OWNS", 1.0);

// Batch linking
await vectorStore.linkBatch([
    { sourceId: 1, targetId: 2, relation: "OWNS", weight: 1.0 },
    { sourceId: 2, targetId: 3, relation: "DEPENDS_ON", weight: 0.95 }
]);

// Declarative edge induction from metadata
await vectorStore.induceEdgesFromMetadata([
    { field: "file_path", ruleType: "path_hierarchy", relation: "CONTAINS" }
]);

// Tribunal anti-hallucination verification
const verification = await vectorStore.verifyGeneration(
    llmAnswer,
    ["1", "2"]
);

📡 Edge-Cloud Handshake

Synchronize episodic memory between local robot/agent and the fleet:

// Handshake hashes to identify memory drift
const digest = await client.getDigest("agent_memories");
await syncWithCloud(digest.state_hash, digest.buckets);

📖 Documentation

• Why Spatial AI? (YARlabs Manifesto)
• HyperspaceDB Official Docs

📄 License

Apache-2.0. Copyright © 2026 YARlabs.