Local, encrypted, provenance-aware memory for AI agents.

How it works

One engine, linked into your process.

1. 1

   Store

   Documents, facts and the embeddings you choose go into one append-only, checksummed log on your disk, encrypted if you turn it on.

2. 2

   Index

   BM25 postings, a vertex-graph vector index and superposed fact vectors are built in the same process. No server, no network call.

3. 3

   Recall

   Hybrid search fuses lexical and vector scores and returns byte offsets into the source; signed provenance records who stored what, and when.

The holographic part is the fact memory: bound facts are superposed into fixed-size vectors and recalled by similarity. Document search is conventional BM25 and cosine.

Quickstart

Store a document and find it again. The five-minute quickstart adds encryption and filters.

# Cargo.toml: Apache-2.0 source from main
cargo add holographic-memory --git https://github.com/writerslogic/holographic-memory
cargo add serde_json anyhow

use holographic_memory::{DocumentInput, HmsCore, SearchOptions};

fn main() -> anyhow::Result<()> {
    // 4,096-dimension store in ./memory; nothing leaves this machine.
    let hms = HmsCore::new(4096, Some("./memory".into()), None)?;
    hms.memorize_document(DocumentInput {
        id: "backups".into(),
        text: "Restore deleted documents from verified backups.".into(),
        source_uri: Some("operations.md".into()),
        version: Some("1".into()),
        metadata: Some(serde_json::json!({ "project": "alpha" })),
        chunk_words: None, overlap_words: None, store_text: None,
        embeddings: None, title_weight: None,
    })?;
    let hits = hms.search_documents("restore backups", &SearchOptions::default())?;
    println!("{} {:?}", hits[0].document_id, hits[0].source_uri);
    Ok(())
}

# hms-server stores vectors you supply; the SDK is its HTTP client
cargo build --release --features server --bin hms-server
HMS_API_KEY=change-me target/release/hms-server --input-dim 384 &
pip install holographic-sdk

from holographic_sdk import HolographicClient

client = HolographicClient(url="http://127.0.0.1:8080",
                           api_key="change-me", tenant_id="demo")
client.add_documents([{"id": "a", "vector": [0.1] * 384,
                       "text": "hello", "metadata": {}}])
print(client.query([0.1] * 384, top_k=1)["matches"][0]["id"])

Open source and plans

Your memory stays on your machines. Cloud keeps your machines in step: encrypted sync across your laptops, desktops and team, a backup you can restore anywhere, and independent verification of the provenance and deletion records your stores sign. We never hold your keys, never see your documents and never run your searches. WritersLogic Holographic Memory Cloud runs its sync, backup and verification on Cloudflare.

• Community

  Available now

  Free

  The open-source engine and server on your own machines.

  • Rust library, Python SDK, C ABI with Go, Java and .NET
  • Hybrid search, fact memory, crash-safe log
  • Optional encryption at rest and signed provenance
  • Support through GitHub
  Get started on GitHub

• Cloud

  In development

  Usage-based

  Sync, backup and verification for the memory you keep locally. Pay for encrypted storage, sync operations and verifications, each with a free monthly allowance.

  • End-to-end encrypted sync (AES-256-GCM, keys only on your devices); concurrent edits converge and conflicts are reported
  • Encrypted backup of each store's full history; restore any device with your sync key
  • Tamper-evident history: reordered, altered or spliced changes are detected on your device
  • Provenance and deletion-record verification with receipts you can check offline
  • Signed audit export; business-hours email support
  Join the waitlist

• Team

  In development

  Monthly plan

  Encrypted sync and verification shared across a team, with usage included each month.

  • Everything in Cloud, with included monthly usage and a larger storage quota
  • Scoped API keys with rotation and instant revocation
  • SSO through your identity provider (OIDC) and a signed, exportable audit log
  • Shared support channel and uptime commitment
  Join the waitlist

• Enterprise

  Talk to us

  Talk to us

  Your keys, your infrastructure, compliance reporting.

  • Annual agreement with custom quotas
  • Bring your own key (KMS / HSM)
  • Transparency log and verification you can run yourself
  • Compliance reports and audit retention
  • Named support contact
  Join the waitlist

The source on main is Apache-2.0. Packages published so far (crate 0.6.1, holographic-sdk 0.1.2 on PyPI) were released under AGPL-3.0-or-later and stay available under those terms.