Exogram is the execution control plane for autonomous AI. It enforces truth, detects contradictions, and ensures every agent action is admissible before execution.

How does Exogram govern AI agents?

Exogram sits between AI reasoning and AI action. It evaluates every proposed action against verified truth, constraints, provenance, and temporal state — and blocks anything that violates them.

Does Exogram train on user data?

No. Exogram does not train models on user data. All ingestion is user-initiated and fully auditable.

Exogram is a governed AI memory layer that gives LLMs persistent, verifiable, encrypted memory. Unlike raw vector databases, Exogram verifies facts, detects contradictions, enforces constraints, and encrypts everything at rest — enabling trust-grade AI memory.

How does Exogram differ from Mem0?

Mem0 stores memories as key-value pairs. Exogram goes further with governed verification (every fact is conflict-checked), zero-trust encryption (Fernet AES per-user), PII scrubbing, a full audit trail, and native MCP integration for Claude Desktop.

How does Exogram differ from Zep?

Zep focuses on conversation history management. Exogram focuses on governed fact memory — each entry is verified for contradictions, encrypted with per-user keys, and stored with full provenance. Exogram also supports MCP, CustomGPT, and Chrome extension interfaces.

Does Exogram encrypt AI memories?

Yes. Exogram uses Fernet (AES-128-CBC + HMAC) encryption with per-user salts derived from a master key. All data is encrypted at rest. PII is automatically detected and scrubbed before storage.

Can I use Exogram with Claude Desktop?

Yes. Exogram ships with a native MCP (Model Context Protocol) server that integrates directly with Claude Desktop. It exposes 5 tools: store_memory, recall, check_fact, search_memories, and health_check.

AI Memory Layer Comparison — 2026

Why Teams Choose Exogram Over Alternatives

Most AI memory tools just store text and retrieve it with vectors. Exogram is the first governed verification layer — every fact is conflict-checked, PII-scrubbed, encrypted, and audit-trailed before it enters your AI's memory.

Start Free Trial Read the Docs

What Makes Exogram Different

Governed Verification

Every fact passes a multi-LLM pipeline: PII scrubbing → constraint checks → conflict detection → encryption → vector indexing. No unverified data enters your memory.

Zero-Trust Encryption

Fernet (AES-128-CBC + HMAC) with per-user salts. Your data is encrypted before storage — we can't read it even if we wanted to. SOC-2 ready out of the box.

Conflict Intelligence

Claude-powered contradiction detection catches "My daughter is Dara" vs "My daughter is Sara" — direct, temporal, and numerical conflicts with confidence scoring.

Feature-by-Feature Comparison

Exogram vs Mem0 vs Zep vs LangMem vs MemGPT vs Pinecone

Feature	Exogram	Mem0	Zep	LangMem	MemGPT	Pinecone
Governance
Conflict Detection (LLM-powered)
Constraint Enforcement
Provenance Audit Trail
Cognitive Decay Modeling
Fact Verification Pipeline
Security
Per-User Encryption (AES/Fernet)
Automatic PII Scrubbing
Zero-Trust Architecture
SOC-2 Readiness
Integration
MCP Server (Claude Desktop)
CustomGPT (ChatGPT Actions)
Chrome Extension
REST API
Multi-LLM Orchestration
Memory
Semantic Vector Search
Governed Ledger (not raw store)
Multi-Namespace Isolation
Versioned Memory Entries

Full support Partial Not available

Head-to-Head Comparisons

Exogram vs Mem0

Memory layer for AI apps

Their strength: Simple key-value store, easy Python SDK, quick integration

Why Exogram wins: Exogram adds verification, encryption, conflict detection, and audit trails. Mem0 stores raw memories — Exogram governs them.

Exogram vs Zep

Long-term memory for AI assistants

Their strength: Good conversation history management, session context

Why Exogram wins: Zep stores conversations. Exogram verifies facts within those conversations. Every claim is conflict-checked and encrypted, not just stored.

Exogram vs LangMem

Memory management for LangChain

Their strength: Deep LangChain integration, entity extraction

Why Exogram wins: LangMem is tied to LangChain. Exogram works with any LLM via MCP, REST API, or CustomGPT — and adds governed verification LangMem lacks.

Exogram vs MemGPT / Letta

Self-editing memory for agents

Their strength: Autonomous memory management, tiered storage

Why Exogram wins: MemGPT lets the AI edit its own memory unsupervised. Exogram requires verification — no fact enters memory without passing conflict checks.

Integration Ecosystem

Exogram works wherever your AI does

MCP (Claude)

Native MCP server

CustomGPT

ChatGPT Actions

Chrome Ext.

Browser capture

REST API

Any platform

Frequently Asked Questions

Ready for Governed AI Memory?

Start free with 50 verified memories. No credit card required. Upgrade when you need unlimited evaluations and enterprise features.

Start Free Read the Protocol (RFC-0001)