amemor-ai®
Post-Quantum Encrypted · Local-First · Open Core

Smart Memory
for AI Agents — works with Claude Code, Codex & local Ollama

Your AI agents forget everything between sessions. amemor-ai gives them persistent, secure, auditable memory — so they get smarter with every interaction.

Already a customer? Log in to your portal →

🔧 Installation, security, or setup? Let your voice assistant and smart-install handle all of it for you →
1,100+ automated tests
Zero external dependencies
Post-quantum encryption
GDPR & ISO 27001 ready

Why amemor-ai?

Everything your AI agents need to remember — securely and organized.

Local-First

Your data stays on your hardware. SQLite-backed, zero cloud dependencies. Runs on everything from Raspberry Pi to enterprise servers.

Learn more →

GDPR & Compliance Ready

Built-in compliance auditor, consent tracking, DPIA generator, PII sanitisation. Ready for ISO 27001, GDPR Article 35, and regulatory audits.

Learn more →

Team Sync & Federation

Share knowledge across agents and machines. Federation Hub syncs memory with role-based access control and approval gates.

Learn more →

Enterprise Security

Post-quantum encryption (ML-KEM-768), key escrow, FIDO2 hardware auth — plus a self-securing host and built-in IT expert.

Learn more →

Cloud PII Anonymization

Personal data is anonymized at write time and again at the LLM proxy — the cloud model only ever sees ciphered placeholders. De-anonymization is local, on-demand and audited, so you use frontier cloud models without exposing real PII.

Learn more →

Voice Assistant

A full-fledged voice assistant with its own separate memory, installable as an end-client of the main shared memory. It listens and answers locally — your voice never has to leave the machine.

Learn more →

Built-in IT & Security Expert

Your own IT and security expert for your computer and your network. It watches, hardens and heals the host, explains problems in plain language, and escalates only what truly needs you.

Learn more →

Verified Knowledge Packs

Curated knowledge sets with verified facts, sources and a digital signature — import them into memory as ready-made expertise. Every fact carries its provenance and verification date, so you always know what your AI believes and why.

Learn more →

Built for your world

One private, auditable AI — shaped to how you actually work.

Up and Running in 60 Seconds

Three steps to persistent AI memory.

1

Install

One command: pip install uaml-memory — that's it. No servers, no Docker, no cloud accounts.

2

Configure

Run uaml init to create your memory database. Choose your modules, set up policies.

3

Your AI Remembers

Your agents now have persistent memory. They learn, recall, and reason across sessions — automatically.

Everything amemor-ai does

Eight capability pillars. Expand any heading for the detail, then open the full page for the deep dive. There is a lot here — that is the point.

🧠Memory & knowledgeThe core: AI that genuinely remembers across sessions, with point-in-time truth.7 live features
🛡️Privacy, security & complianceThe buying criterion for regulated industries — AI you can put in front of an auditor.5 live features
🌐Distributed team & orchestrationMany agents, many nodes, no central point of failure.6 live features
💬Studio — the human web UIThe self-hosted EU alternative to cloud AI workspaces — same UX, data never leaves.6 live features
📦Smart Install — distribution & opsA runtime that installs and heals itself — the OEM/MSP automation advantage.5 live features
🎙️Voice assistant (“Adam/Eva”)A computer that greets you and walks you through setup like a colleague. Adam and Eva are just the defaults — name your assistant anything, pick any voice.1 live features
📜Registrar — legal records & long-term validityAI that doesn't just remember — it produces court-valid, self-maintaining evidence.7 live features
🏗️Commercial & platformA proprietary runtime for building bespoke vertical AI products.0 live features

Editions & partnership

Enterprise

Individual offer — pricing is agreed in a meeting.
For larger deployments: memory across multiple machines, a compliance pack.
Talk to us

Custom

Individual offer — pricing is agreed in a meeting.
For larger deployments: federation.
Talk to us

OEM & white-label

Individual offer — pricing is agreed in a meeting.
For larger deployments: OEM and white-label.
Talk to us

For individuals and small teams: uaml-memory.com →

Frequently Asked Questions

What is amemor-ai?

UAML (Universal Agent Memory Layer) is a memory infrastructure for AI agents. It gives your agents persistent, structured memory with 5 types (episodic, semantic, procedural, reasoning, associative) — modeled after human cognition. Your agents remember what they've learned, decisions they've made, and context from past interactions.

How does licensing work?

amemor-ai is proprietary software by GLG, a.s. The Community edition is free for personal, non-commercial use — without the right to modify the software, resell it, or build it into your own products. Commercial use requires a paid subscription. Paid tiers unlock additional features and include professional support. Licenses are per organization; billing is monthly, semi-annual or annual.

Can I try it for free?

Absolutely. The Community tier is fully functional — not a trial, not a demo. You get the full memory engine, post-quantum encryption, CLI, and Python API. Install with pip install uaml-memory and start building immediately. The free tier does not include the Voice assistant or the fully automatic smart install.

Where is my data stored?

UAML is local-first. Your data stays on your hardware in a SQLite database. There are zero cloud dependencies — no data ever leaves your machine unless you explicitly configure federation or export. You own your data completely.

Is UAML GDPR compliant?

Yes. UAML includes built-in GDPR tools: consent tracking, right to erasure (selective purge), data portability (export), PII auto-detection and sanitization, DPIA generator, and breach notification register. It's designed for organizations that need to demonstrate compliance. Learn more →

What makes the encryption "post-quantum"?

UAML uses ML-KEM-768 (formerly CRYSTALS-Kyber), a lattice-based key encapsulation mechanism standardized by NIST as FIPS 203. This protects your data against both classical and future quantum computer attacks. Key escrow with Shamir's secret sharing provides recovery without single points of failure.

Applications — case studies

Real deployments, measured results, verifiable proof — no invented customers.
🧠Case study: 77% of entities lost — then 100% recoveredPublished research on production data: why context quality beats model sizeRead the case study →📧Case study: an inbox triaged by disposable AI agentsThe most hostile input on the internet, handled in an isolated VM — months in productionRead the case study →📜Case study: documents that prove themselves — for decadesQualified seal + self-renewing timestamps: INTACT:TRUSTED in any readerRead the case study →🎙️Case study: the computer that introduces itselfVoice onboarding during install — fully offline, even on an aging workstation GPURead the case study →🌐Case study: this website translated itselfEvery language you can switch to here was produced by the product's own pipelineRead the case study →