01
decision2m
Usar bloqueos consultivos en PostgreSQL
Fuente: Nelson · architecture.md
SELF-HOSTED AI MEMORY · AVAILABLE NOW
Self-Hosted AI Memory in Dublin
Hosted or self-hosted, same MCP/REST protocol. Start hosted in 2 minutes, move to your infra later with a config change — zero code change.
Stop shipping code context to a black box. MemHub is self-hosted AI memory on your PostgreSQL 16 + pgvector — Docker or Easypanel, you run the migrations, you own the data.
N
Nelson · Founder, MemHub at Simplex.lat
Simplex.lat · Quito, Ecuador ·
Published by MemHub at Simplex.lat. Product limits come from the current plans; validate capabilities in your own pilot.
Local-first · OfflineSin pérdida de contextoSetup en 2 min
Start free — get briefed in 2 min
Free for 1 project · No card · Self-host or we host · No external API
acme/self-hosted-ai-memory
synchronizedcontext/main
memorias 41
fijadas 6
autores 9
BRIEFING FOR NEW SESSION
What changed since your last pull
02
pattern18m
Los errores de API siguen RFC 9457
Fuente: Maya · PR #184
03
bugfix1h
Los refresh tokens rotan atómicamente
Fuente: Codex CLI · auth/session.ts
Claude Code
Cursor
Codex CLI
Definition
What is Self-Hosted AI Memory in Dublin?
Self-Hosted AI Memory is how engineering teams preserve the why behind the code — rejected alternatives, pinned invariants and recent fixes — as searchable, versioned entries. In Dublin, MemHub delivers that memory over MCP with incremental briefings, offline support and auditable authorship.
Team plan at a glance
$29
Team per workspace per month (USD)
Team plan price; review current terms before subscribing. This is not a savings estimate.
According to MemHub pricing →5
Seats included in Team
Share decisions with authorized teammates. Shared memory is not a replacement for live collaborative chat.
According to MemHub pricing →20
Projects included in Team
Configure project scope and test what each user can retrieve before connecting sensitive repositories.
According to MemHub pricing →What changes the first week
In the first week with Self-Hosted AI Memory in Dublin, sessions stop re-proposing settled decisions because pinned invariants outrank noise during ranked retrieval. According to Authoritas, AI Overviews cut traditional clicks by 58%, so being the cited, extractable source inside the briefing matters more than dumping more tokens.
Standups shrink to real blockers, reviews surface the original rationale in one query, and agents respect hard constraints automatically. Because briefings are deltas, the archive stays a query away rather than a stowaway in every prompt.
Why Dublin teams need Self-Hosted AI Memory
Self-Hosted AI Memory in Dublin helps engineering teams ship faster because every AI coding session opens pre-briefed with only what changed since the last pull, not a full dump. According to Search Engine Journal, 45% of searches now show an AI Overview, so concise, cited memory answers also travel beyond the IDE into generative results.
Distributed squads stay aligned across time zones, new hires ramp in days from searchable precedent, and mixed toolchains stop fragmenting knowledge. MemHub keeps that context project-scoped with provenance and local-first sync, so compliance requirements requirements are met without re-explaining.
THE MECHANISM
Write once. Every session remembers.
Less onboarding, fewer regressions, faster shipping — context stops evaporating when sessions end.
01
You keep coding
Decisions auto-captured where you work — terminal, PR, editor. No extra docs, no context-switching.
02
MemHub does the plumbing
Local SQLite outbox first, then project-scoped sync with provenance (who, when, why). Works offline.
03
Next session starts ahead
Every teammate and agent opens pre-briefed with only what changed since their last pull.
UNIVERSAL COMPATIBILITY
Plugs into any CLI, IDE, or AI coding agent.
If it speaks MCP or reads a JSON config, MemHub plugs in. One command wires your editors and terminals without friction.
- Claude Code
- Cursor
- Windsurf
- Cline
- Codex CLI (GPT)
- GitHub Copilot
- Google Antigravity
- Gemini CLI
- Aider
- OpenCode
- CI bots via REST
WHAT YOU GET
Self-Hosted AI Memory: core benefits and capabilities
01
Move without rewriting
Hosted and self-hosted speak identical MCP/REST — switch is config, not code.
02
Stay inside your boundary
One Docker Compose / Easypanel file — database never phones home, data never leaves.
03
Production posture day one
RLS policies, NOBYPASSRLS roles, transaction-local tenants — in the migrations, not a todo.
04
Answer the auditor fast
Append-only audit table with actor/reason/metadata — demos that bore the DPO (in a good way).
05
AI without the API bill
Start with FTS; flip on local Ollama for vectors — no external embedding vendor needed.
06
Backups you already run
Plain PostgreSQL — pg_dump, replicas, PITR fit your existing continuity plan.
The honest comparison.
No feature-matrix gymnastics — just what changes the week you adopt MemHub.
| Feature | MemHub | Mem0 | Supermemory | Ideal For |
|---|---|---|---|---|
| Scope | Organization → Project → Member, shared across every agent | Per-user graph, not team-scoped | Per-user vault with sharing on request | Teams that need one truth across tools |
| Storage | PostgreSQL 16 + pgvector, versioned & queryable | Vector graph on managed infra | Hosted cloud store, limited export | Teams that need SQL, backups and audit |
| Tenancy & RLS | Row-level security inside the DB, enforced per project | Application-level isolation | Workspace isolation, no RLS | GDPR / SOC-2 reviews |
| Self-host | Docker/Easypanel recipe included, same MCP + REST | Open-source self-host possible, different ops | Cloud-only | Data-residency and on-prem requirements |
| Offline & sync | Local SQLite outbox, incremental deltas by sequence | Online-first sync | Online-first, no local outbox | Flaky networks, planes and VPN gaps |
| Pricing model | Community free (1 project) · Team $29 · self-host unlimited | Free tier + usage billing | Free tier + pay-per-seat | Trying without procurement |
Keep your stack. Add shared memory.
One command for Claude Code, Cursor, Codex, Windsurf, Cline or any MCP client. No new editor. Setup in 2 minutes, works offline.
memhub-cli · preview
Click tabs to simulateSimulate CLI
$
memhub save --type decision "Usar bloqueos consultivos en Postgres" -m "Evita colisiones concurrentes en el outbox"stdout · [mem_save]12ms
[detect] Contexto resuelto: acme/payments-api
[sqlite] Guardando en outbox local (~/.memhub/outbox.db)...
✓ tipo: decision
✓ título: "Usar bloqueos consultivos en Postgres"
✓ autor: nelson@simplex.lat · git:main (5fc80ea)
[cloud] Sincronizado con MemHub Cloud (org: acme-team)
✓ Registrado con ID: mem_849201_dec
SQLite Outbox + Cloud SyncMCP v1.0
Questions, answered straight.
Is the hosted option still private?
Hosted memory is isolated per organization with row-level security, encrypted transport, and hashed credentials — but self-hosting remains the strongest guarantee because no third party operates the storage. The result is fewer regressions, faster onboarding and prompts that carry current, ranked context rather than stale boilerplate repeated across every agent.
What do I need to self-host?
A host with Docker (or Easypanel) and a PostgreSQL 16 database with the pgvector extension. Migrations and seeds are one command each. The result is fewer regressions, faster onboarding and prompts that carry current, ranked context rather than stale boilerplate repeated across every agent.
What happens when our connection drops?
Nothing interrupts the work. Writes commit to the local SQLite outbox and synchronize automatically when the shared service is reachable again. The result is fewer regressions, faster onboarding and prompts that carry current, ranked context rather than stale boilerplate repeated across every agent.
Can we try it without sending data anywhere?
Yes — the local engine works fully offline, and the self-hosted stack keeps even the shared layer on your own PostgreSQL. The result is fewer regressions, faster onboarding and prompts that carry current, ranked context rather than stale boilerplate repeated across every agent.
How much does it cost?
A Community tier is free, and the Team plan adds multi-project contexts, seats, and higher synchronization limits. Self-hosting is unlimited. The result is fewer regressions, faster onboarding and prompts that carry current, ranked context rather than stale boilerplate repeated across every agent.
MORE LOCATIONS
Self-Hosted AI Memory in San FranciscoSelf-Hosted AI Memory in New York CitySelf-Hosted AI Memory in AustinSelf-Hosted AI Memory in SeattleSelf-Hosted AI Memory in BostonSelf-Hosted AI Memory in ChicagoSelf-Hosted AI Memory in DenverSelf-Hosted AI Memory in Los AngelesSelf-Hosted AI Memory in AtlantaSelf-Hosted AI Memory in MiamiSelf-Hosted AI Memory in DallasSelf-Hosted AI Memory in San Diego
EXPLORE OTHER USE CASES
AI Agent Memory
Give every AI coding agent a persistent, team-wide memory.
MCP Memory Server
A hosted MCP memory server your whole team connects to in one command.
Claude Code Memory
Claude Code memory that survives sessions and reaches the whole team.
Cursor Memory
Carry Cursor context across sessions, machines, and teammates.
Codex CLI Memory
Persistent context for Codex CLI — personal today, shared tomorrow.
Cross-Agent Memory
One memory behind Claude Code, Cursor, Codex CLI, and Antigravity.
Developer Knowledge Base
A living knowledge base that writes itself while your team codes.
Coding Agent Context
Context engineering for coding agents — brief, don’t dump.
Engineering Decision Log
An ADR habit your team will actually keep — logged by the agents doing the work.
Agent Memory Database
A real database behind agent memory — PostgreSQL, pgvector, row-level security.
Windsurf Memory
Persistent Windsurf context shared with every other agent on the team.
Cline Memory
Give Cline a memory that survives tasks, machines, and teammates.
Copilot Team Memory
What Copilot’s custom instructions can’t hold, MemHub remembers.
Antigravity & Gemini Memory
Persistent context for Google Antigravity, Gemini CLI, and the agents around them.
THE NEXT SESSION STARTS HERE — 2 MIN SETUP