Enterprise Security
How CortexDB approaches security for production deployments. We secure the platform using PASETO v4 bearer authentication, hierarchical scopes, and 4-tier capability stacks.
The notes
What we wrote down while building it.
Benchmark runs with their cost and methodology attached, the architecture written out in full, the head-to-head comparisons we would have wanted before choosing a memory layer, and the tutorials for wiring one into an agent.
16 posts
Deployment Profiles
How CortexDB uses deployment profiles to gate enterprise controls. We support four profiles—startup, growth, enterprise, and custom—to match your operational maturity.
Benchmark Results
How CortexDB performs on long-term memory benchmarks. We report 93.8% on LongMemEval-S using our lossless event-sourced architecture and 4-channel hybrid retrieval.
Backups and Disaster Recovery
How CortexDB handles backups and disaster recovery. We utilize RocksDB checkpoints, object-store upload targets, and WAL archival to enable point-in-time recovery.
Audit Trail
How CortexDB logs every API operation for compliance, security monitoring, and debugging. The system captures structured fields, supports SIEM export, and ensures verifiable deletion.
CortexDB Architecture
How CortexDB leverages a lossless event-sourced architecture, async knowledge graph enrichment, and 4-channel hybrid retrieval for AI agent memory.
Run CortexDB on Your Own Hardware in Under Five Minutes
v0.5.1 ships CortexDB as a Docker image and stripped Linux binaries — the same engine that powers CortexDB Cloud, now runnable inside your own VPC, your homelab, or your laptop.
CortexDB V1: The Experience Layer for AI Agents
V1 of CortexDB is live: 93.8% on LongMemEval-S (beating Mem0), 86.9% on LoCoMo, the five-layer memory model, hierarchical scopes, PASETO auth, and 31 integrations plus 16 connectors — built in 8 weeks.
CortexDB vs the Competition: A Production-Scale Memory Benchmark
A controlled internal comparison of CortexDB against an LLM-rewrite-on-write baseline across five production-scale scenarios. CortexDB reports 84.2% retrieval accuracy against the baseline's 31.8% at 1/337th the daily write-path cost.
CortexDB vs Mem0: An Architectural Comparison for AI Agent Memory
CortexDB and Mem0 take different approaches to AI agent memory. CortexDB ships lossless event-sourced memory. Mem0 defaults to LLM-extraction on writes.
CortexDB vs Zep: Architecture, Retrieval, and Multi-Tenancy Compared
CortexDB ships lossless event-sourced memory with 4-channel hybrid retrieval. Zep builds temporal knowledge graphs from chat and business data. Architecture, knowledge graph, and multi-tenancy compared.
93.8% on LongMemEval-S: How CortexDB's Retrieval Stack Stacks Up
A single-run 93.8% on LongMemEval-S (469 of 500), running against the production CortexDB server with no in-memory shortcut. Pipeline walkthrough, per-category scores, leaderboard, and cost and latency analysis.
How to Add Memory to Your Hermes Agent
A walkthrough of using CortexDB as a long-term memory provider for NousResearch's Hermes Agent — with persistent recall, background writes, and prefetching.
Give Your OpenClaw Agent a Real Memory
OpenClaw turns Claude into a desktop coworker with eyes and hands. CortexDB gives that coworker a memory that survives restarts, machines, and weeks.
Building an AI Companion with Perfect Memory
A step-by-step guide to building an AI companion that remembers everything — preferences, past conversations, decisions, and context — using OpenAI and CortexDB.
Introducing CortexDB: The Long-Term Memory Layer for AI Systems
Why we built an event-sourced memory layer for AI agents — durable memory, connected context, 31 integrations, 16 connectors, and production-ready scalability.
Nothing here yet. No posts in that category.
Next
The record is a separate document.
What shipped and when lives on the changelog — sixty-three releases with their full notes. This page is the thinking; that one is the log.