One memory. Every mind you work with.
One memory family, three doors: personal knowledge, team code memory, and machines that remember where they have been. Persistent, shared memory in the Open Knowledge Format — yours to export, forever.
Your AI should not start from zero every morning. SynapseTree remembers what matters — your stories, your codebase, your machines’ missions — and makes that memory available wherever you work next. No coding required to start. No lock-in when you leave.
Everything you teach your AI evaporates.
Today’s AI tools are brilliant and forgetful. You end up paying — in time, in tokens, in repeated explanations — to teach the same lessons again and again.
It forgets between sessions
Every new conversation starts from scratch. The background you built up yesterday has to be rebuilt today, by hand.
It forgets when context fills
Long projects overflow any single conversation. The early decisions — often the most important ones — are the first to fall out.
It forgets across devices
Notes on your phone, drafts on your tablet, work on your desktop: three islands, no bridges. Your memory should sync like your photos do.
It forgets across your team
Your assistant learns your codebase or your world. Everyone else's assistant starts dark. Shared work deserves shared memory.
One memory family, three doors.
SynapseTree is not three separate products. It is one memory, reached through whichever door fits the moment. Start at any door — what you save in one place is remembered in all of them.
Curated Journal
Your private memory engine, in your pocket. An offline-first mobile app that captures and organizes your knowledge on your phone — private by default, useful with no connection at all. Free and open source (MIT).
Curated Thoughts
Your desktop AI companion. A bigger canvas for thinking, writing, and working with AI over everything your memory holds.
SynapseTree
The cloud memory and sync hub. The shared brain that connects your devices, your team, your coding agents, and your machines — securely, and always exportable.
For software teams — the multiplayer memory layer for AI coding agents.
Stop re-indexing codebases and burning tokens on context you’ve already taught your AI. Give Claude Code, Cursor, and Windsurf a shared, persistent brain.
No more re-indexing tax
Your agent stops re-learning the same codebase every session and stops burning tokens on context you already paid for once.
Decisions that survive
Architectural decision records and the reasoning behind them persist in the team graph, even when a context window fills.
Multiplayer by default
Patterns one developer's agent learns sync to the team graph, so every teammate's agent starts informed.
Fast context hydration
Pre-computed knowledge graphs load in microseconds — no waiting on massive cloud vector queries.
How to get started in 3 steps
Install the MCP server
npm install -g @equationalapplications/mcp-server
Generate your API key
Create a free account, then link your agent to its dedicated cloud database: synapsetree-mcp auth --key stree_…
Code — agents sync automatically
Your agent reads and writes structured knowledge to the isolated cloud database. Context is saved forever.
For robotics — machines that remember where they have been.
A drone or rover with no memory treats every mission as its first. Synapse Core connects your machines to the same memory family, so they remember places, missions, and lessons — and share that knowledge across your fleet.
Synapse Core Lite (in development)
A low-cost safety brick that pairs with your phone as the brain, for getting a first vehicle remembering missions with your phone in the loop.
Synapse Core F (Field) (in development)
Our lead product: a dual-chip brick with its own on-board episodic memory — no phone required, for field vehicles that must keep logging and recalling missions on their own.
Synapse Core (in development)
The full compute module with on-board knowledge graphs — for autonomous offline operation and fleet-scale deployments.
A spatial ontology for robots
Places, routes, geofences, and mission history your machines can reason over — not just log files.
Safety stays local
A deterministic on-board reflex guards every machine; the AI proposes, the safety layer disposes. Memory never overrides physics.
Fleet graph consolidation
Devices buffer raw telemetry offline and sync to SynapseTree upon reconnecting to build shared fleet intelligence.
Inference is your choice.
We do not charge for AI compute, and we never mark up tokens on our base plans. You pick how your AI thinks, per app and per task.
Local-first
Models run on your own device. Fast, private, and free — the default wherever it can work.
Bring your own endpoint
Point SynapseTree at your own inference endpoint and keep full control.
Bring your own keys
Use your own OpenAI, OpenRouter, or other provider keys. Your keys, your bill, no markup — ever.
Managed inference (coming soon)
Prefer not to manage keys? Managed AWS Bedrock inference (Claude, Nova) through us is planned as an opt-in paid add-on. It is not available yet — the bring-your-own path is here today and never goes away.
Two AI channels, tuned independently.
Good memory needs two different kinds of AI: one that reasons, and one that finds. SynapseTree gives each its own channel — a language-model (LLM) interface for reasoning and writing, and a separate embedding interface for vector search and recall. Configure each independently: a local model for one, a cloud key for the other, whatever fits. Your memory’s search quality and your AI’s reasoning quality no longer rise and fall together.
OKF-first: your memory is never locked in.
Everything SynapseTree stores is kept in the Open Knowledge Format (OKF) — an open, portable representation of knowledge. Your memory is a plain, readable file, not a hostage. Export it anytime, use it with other tools, or run the open-source engine yourself. If we ever disappeared, your second brain would not.
Private by design.
Local-first
Your memory lives on your devices first; the cloud is a sync hub, not the only copy.
Explicit consent
Nothing fails over to cloud inference silently. You choose, per channel, what may leave your device.
Absolute data sovereignty
Every workspace gets its own isolated database. No shared tables, no cross-tenant access, and we never train models on your data.
Export anytime
Download your memory and use it offline, forever.
Shared trust & core commitments.
You pay for memory, not compute
Base plans use your own API keys with zero token markup.
No lock-in (OKF-first)
Everything is stored in the Open Knowledge Format (OKF). Export your plain files and run them offline with open-source tools anytime.
Absolute data sovereignty
Dedicated, database-per-tenant isolation with zero cross-tenant access and no training on user data.
Pricing
We do not charge for AI compute. You only pay for active databases and storage — bring your own inference, or add managed inference later if you want it.
Portability is free forever. Export and re-import your memory as Open Knowledge Format files — on any plan, with no export fees or lock-in. Per-export size caps apply and are disclosed beside the export control and in the Product Terms. That is the promise that never changes.
For your personal second brain, or a single developer syncing a codebase across devices.
- 1 cloud database
- 50 MB storage
- Isolated database, export anytime
For writers with a series to track, freelancers, and indie hackers with several projects on the go.
- Up to 5 cloud databases
- Automated nightly backups
- Cross-device sync, priority support
For teams — human, agent, or fleet — sharing one memory graph.
- Up to 20 cloud databases
- Team roles and access control (RBAC)
- Single-writer queues, multiplayer write protection
Frequently asked questions
Do I need to code to use SynapseTree?
No. Curated Journal and Curated Thoughts are ordinary apps — install, save, and ask. The coding-agent tools are there for developers who want them; they are a door, not a toll booth.
What about robots?
Synapse Core hardware (Lite, F (Field), and Core tiers — all in development) connects drones and rovers to the same memory family, so machines remember places, missions, and fleet knowledge. Robotics is an optional door into the same memory — your personal and team knowledge stays separate unless you choose to share it.
What is OKF?
The Open Knowledge Format: an open, portable way to store knowledge as plain, readable files. Because SynapseTree is OKF-first, your memory works with other tools and outlives any single app — including ours.
Do you charge for AI or token usage?
No. On base plans, SynapseTree is bring-your-own-inference: your apps and agents use your own endpoint or your own provider keys, and there is no token markup. Managed AWS Bedrock inference is planned as an optional paid add-on for later; it is not offered yet.
Can I self-host?
Yes. The core engine and Curated Journal are open source, MIT licensed, and free — a local database is just your own file. The hosted service exists for cloud sync, team sharing, and multi-device persistence.
How is my data isolated?
Strictly database-per-tenant: your workspace gets its own dedicated database in its own location. There are no shared tables and no cross-tenant query path, and we never train models on your data.
Can I leave?
Anytime. Download your memory in the open OKF format and use it offline with the open-source engine. Your memory is a plain file, not a hostage.
Start remembering.
Pick your door. Writers and thinkers start with Curated Journal. Teams start with the MCP server. Builders start with Synapse Core. Wherever you begin, it is the same memory underneath — yours, portable, and shared with exactly who you choose.
Get Your Free Cloud Database