Verifiable memory · as a service

Total recall
for every agent.

Drop-in memory infrastructure that retrieves the right fact, routes every query to the right mode, and grounds every answer in provenance.

62.19 on MemoryAgentBench· ~100ms retrieval· SOC 2 · HIPAA· same API, any deployment
Deployed across
financehealthcareagent platformssupport automation
62.19MemoryAgentBench
above the 49.6 ceiling
Why teams choose SupraCognition

The memory layer for agents that never forget.

☁ Retrieve, don't recompute

Stop paying twice

Agents pull a verified fact in ~100ms instead of re-deriving it over ~30 seconds. The redundant work you stop doing is the line item that moves this quarter.

◇ Routed, not one-size-fits-all

The right mode, every query

A frozen sub-50ms classifier routes each query to one of five cognitive modes — trivial lookups go fast and cheap, high-stakes recall goes deep.

⛓ Provenance, not promises

An answer you can audit

Every recall returns with confidence, sources, and a tamper-evident hash. A record of what was known, and when — audited, not asserted.

Your agents are being re-billed for forgetting

Every agent re-derives the same answer. Alone.

Behind the firewall the waste is worse than on the open web: one rep's agent researches a customer Monday, another's repeats it Thursday, legal re-derives the same clause. The same fact, re-derived and re-billed, team after team.

0%
of enterprise AI users made a major decision on hallucinated output (Deloitte)
0%
hallucination reduction when answers are grounded in retrieved facts
~100ms
to retrieve a verified fact, versus ~30 seconds to recompute it
0×
faster than re-derivation, per identical answer, across your fleet
Supra Cognitive Modes

One classifier. Five modes.

Forcing every question through one retrieval strategy is like using one tool for every job in the workshop. We read each query and route it to the strategy it actually needs.

Single-fact lookup
A key, a lookup, near-instant. “What's their renewal date?”
Time-anchored lookup
A question about time. “What changed since last quarter?”
Long-form synthesis
Gather scattered fragments and reason across them.
Latest-version resolution
When facts conflict over time, surface the one in force.
Pattern from examples
Infer the answer from the shape of many prior cases.
The memory trilemma

You can't pin
all three.

Accuracy, latency, cost — push toward one corner and you give ground on another. So we don't pick one point and live there; the classifier picks a different point for every query.

Accuracy
58%
Latency
2.1s
Cost
2,550tok
Drag the node. Try to turn all three green — you can't.
Routing is pure upside. A strong answer is the floor.
The benchmark that isn't saturated

We lead MemoryAgentBench.

LoCoMo and LongMemEval are saturated — every serious system clusters at the top. MemoryAgentBench is adversarial and multi-session; on multi-hop conflict resolution, most systems collapse below 7%.

MemoryAgentBench — overall score · higher is better
SupraCognition
62.19
Best result in the paper
49.6
Typical agentic-memory system
~37
MemoryAgentBench 62.19 · the hard one LoCoMo 84.86 · saturated LongMemEval 86.0 · saturated

How we report. Backbones vary across systems; our 62.19 exceeds the highest overall result in the paper (49.6). LoCoMo and LongMemEval are saturated, so we report our own measured figures rather than disputed self-reported numbers.

How it works

From a question to a verified answer.

01 · Classify

Route

A frozen sub-50ms classifier reads intent and picks the cognitive mode.

02 · Retrieve

Fuse

Four retrievers plus a version-aware graph select the candidates.

03 · Score

Rank

MemoryRank weights by reliability and recency; stale facts decay.

04 · Synthesize

Compose

A synthesis pass at temperature zero composes the answer.

05 · Anchor

Prove

Confidence and provenance return; the hash anchors to the audit trail.

Done well once.
Audited, not asserted.

No system can declare a fact true — so we don't. What gets anchored is compact and tamper-evident: the hash, the timestamp, and the trail of how a claim's confidence evolved. Run it as a private audit chain, or anchor to Supra L1.

On-chain anchor · private or Supra L1
Knowledge parcelEnterprise · rate policy
Content hash0x0000…0000
Timestamp
anchored · audited, not “true”
Deploy it your way

Hosted, private, or in your own cloud.

Same SDK and API across every deployment — start managed, move private later, rewrite nothing.

☁ Managed cloud

Hosted & managed

Fully managed, autoscaling memory service. The fastest path to production — usage-based, 99.9% SLA, zero infrastructure to run.

Start in minutes →
⌂ Private

In your boundary

Deploy into your VPC, on-prem, or fully air-gapped. BYOK encryption and a private provenance anchor; data never leaves your environment.

For regulated workloads →
⧉ API & SDK

Drop-in integration

Python and Node SDKs, a REST API, and framework plugins. Add verifiable memory to an existing stack without rebuilding your pipeline.

Build today →
CapabilityManaged cloudPrivateAPI & SDK
Classifier-routed retrieval
Hash-chained provenanceprivate anchor
Autoscaling & 99.9% SLAyour infra
BYOK · data residency
Air-gapped option
SSO/SAML · dedicated supportenterpriseenterprise
SOC 2 Type II HIPAA BYOK Hash-chained audit SSO / SAML · RBAC Data residency
Pricing

Start free. Scale on usage. Talk to us for enterprise.

Developer
$0 / mo
For prototyping verifiable memory into an agent.
  • Managed cloud, generous free tier
  • Python & Node SDKs, REST API
  • Classifier-routed retrieval
  • Provenance on every recall
  • Community support
Get API key
Most teams
Team
Usage · metered
For agents in production that need scale and visibility.
  • Everything in Developer
  • Autoscaling & 99.9% SLA
  • Full audit log & observability
  • RBAC and project workspaces
  • Priority support
Start with Team
Enterprise
Custom
Managed or private — for regulated, high-stakes workloads.
  • Private / VPC / air-gapped deploy
  • BYOK, residency, SOC 2, HIPAA
  • SSO / SAML, custom SLA
  • Dedicated solutions engineer
  • Benchmark pilot on your data
Contact sales

Give your agents pristine recall.
Your unfair advantage.

Book a 30-minute call