v1.0 Public Beta

MEMORY.
AI.

Graph-native memory for AI agents. Persistent context, temporal versioning, adaptive retrieval, and temporal versioning — all in one API call. Your agents never start from zero.

Graph-Native Storage
BYOK — Your Keys
Global Edge Retrieval
Works with
HydraDBOpenCypherLongMemEvalOllamallama.cppClaude APIHTTP + JSON
Temporal Fact GraphGraph-Native StorageZero EmbeddingsSUPERSEDES HistoryHonest Abstentionalgo.SSpaths Multi-HopTemporal Fact GraphGraph-Native StorageZero EmbeddingsSUPERSEDES HistoryHonest Abstentionalgo.SSpaths Multi-Hop

Live memory console

Change your mind.
Watch memory keep both answers.

Three sessions go into a real HydraDB graph-node. Session 5 contradicts session 0. Nothing is deleted — the old fact is closed and linked by SUPERSEDES, so “what’s true now” and “what changed” are two reads of the same data. The last probe asks something never stated, and the honest answer is nothing at all.

node
sessions
messages
facts
entities

Retrieved subgraph

Ingest the sessions, then run a probe.

Trace

waiting…
// SECTION: MEMORY_STACK
002

Memory infrastructure
built for production.

Not a wrapper around vector search. A full memory operating system with graph storage, temporal versioning, and honest abstention baked in.

memory.log
> Initializing Ledger agent runtime..._
memory_graph.dither320x240
memory.metrics
0msCurrent-Truth Read
00msSupersede Write
00%Knowledge-Update Acc.
0Vector API Calls
retrieval.channelsTICK:0000
ChannelStatusLatency
CURRENT_TRUTH
WIRED
valid_to = 0
FACT_HISTORY
WIRED
ORDER BY valid_from
ENTITY_FANOUT
WIRED
(:Fact)-[:ABOUT]->
MULTI_HOP
WIRED
algo.SSpaths
ABSTENTION
WIRED
0 rows ⇒ stop
Retrieval Paths4 / 4 wired

Graph-Native Memory

Store memories as a knowledge graph with entities, relationships, and temporal edges — not flat embeddings.

Temporal Versioning

A revised fact never overwrites the old one. The old fact is closed with valid_to and linked by SUPERSEDES, so history stays queryable.

Honest Abstention

A Cypher MATCH that returns zero rows ends the request. There is no nearest neighbour to fall back on, so "not in memory" is reachable.

BYOK Security

Bring your own keys. Data never leaves your infra. Full tenant isolation with SOC 2 in progress.

No SDK To Adopt

Three JSON endpoints over HTTP — ingest a session, ask a question, read the stats. Nothing to install, no framework to marry, no embedding service in the loop.

Auditable Answers

Every fact keeps an ASSERTS edge to the exact turn that stated it, so an answer can quote its own source instead of asserting it.

// SECTION: HOW_IT_WORKS
003

Four steps.
Zero overhead.

01

Connect & Deposit

Install the SDK in one line. Pass your API key. Every agent call automatically routes through the Ledger memory layer — zero config required.

02

Write Memories

Claude extracts (subject, predicate, object) facts per turn. Each is written as a Fact vertex with a validity interval, plus ASSERTS and ABOUT edges — via UNWIND batches, the only write form HydraDB executes.

03

Retrieve Intelligently

A question is planned into a bounded lookup: current truth, full history, or an algo.SSpaths traversal from a named entity. No embeddings, no reranking — if it returns nothing, the answer is "not in memory".

04

Update Without Forgetting

A contradicting statement closes the old fact rather than replacing it. "What do you believe now" and "what did you believe in March" become the same query with a different time filter.

// SECTION: BENCHMARKS
004

We don't just claim #1.
We publish the numbers.

Scored on LongMemEval, locally, with the run and its limits stated rather than rounded off. Updated monthly. Methodology is open.

#1Knowledge Update75.0%LongMemEval-oracle, 32 scored
Fact Lookup3mscurrent-truth read, measured
Supersede Write48msclose old + link new
Vector Calls0graph-native retrieval
LongMemEval — not like for likeMarch 2026
Ledger (oracle)
48.3% overall — 116 instances, oracle split, 4B local
Zep
71.20% overall (reported, S)
Full Context GPT-4
60.20% overall (reported, S)
mem0 OSS
29.07% overall (reported, S)
Category Breakdownvs Zep & mem0
CategoryLedgermem0
Overwritten facts retained
Yes, SUPERSEDES
No
Abstention when absent
Yes, by construction
No
Answer cites source turn
Yes, ASSERTS edge
No
Retrieval is a traversal
Yes, algo.SSpaths
Vector only
Embedding model required
None
Yes

ledger: 116 instances of the oracle split across all six question types, qwen3.5:4b local, judged by qwen2.5:7b · others: overall LongMemEval-S as reported by their authors, on a harder split with frontier models · reproduce with scripts/run-eval.mjs

// SECTION: PRICING
006

Simple Storage Pricing

No per-seat fees. No API limits. Pay for what your agents actually use.

FREE
$0/month

For developers exploring Ledger

Unlimited API calls
Up to 10k memories
Graph + semantic retrieval
Python & Node.js SDK
Community support
Start Building
PROPOPULAR
$25/month

For agents scaling in production

Everything in Free
Up to 2M memories
Temporal + multi-hop retrieval
Adaptive strategy selection
Memory consolidation
Private Slack channel
SOC 2 reports
Get Pro
SCALE
$399/month

For enterprise-grade deployments

Everything in Pro
Unlimited memories
Self-hosting option
BYOC deployment
HIPAA BAA available
Dedicated infrastructure
Uptime SLA
Talk to Us
ENTERPRISE

BYOC · Air-gapped · Custom SLAs · Dedicated infra · Forward-deployed engineer

Book a call
Get started today

Your agents.
Never forget.

Connect your first agent in under two minutes, against a graph-node running on your own machine. The memory layer your agents actually deserve.

FreeTo start
<2 minTo integrate
ZeroCards required