Product · ThoughtOS

Stop Prompting
From Scratch

ThoughtOS is your second brain for AI. It turns scattered chats, notes, preferences, and decisions into a living memory system your assistant can actually use. Less repetition, better follow-through, and an AI that actually grows with you.

The Problem

Smart in the moment. Forgetful over time.

Your goals, style, relationships, open loops, and hard-won insights disappear between sessions. Every conversation is a blank slate.

01

Constant Repetition

You’ve explained your preferences, your team, your architecture a dozen times. Each new session, you start from zero.

02

Platform Lock-In

Knowledge trapped in one app. Switch from ChatGPT to Claude? Everything you’ve taught it stays behind.

03

Lost Decisions

Your AI helped you decide on an approach last week. This week, it doesn’t know the decision exists.

One Brain, Every AI

One second brain.
Every AI. Always yours.

We built ThoughtOS because the AI industry has a memory problem. Every platform creates a walled garden around your knowledge. You invest weeks building context with one tool — then a better model drops and you start from zero.

ThoughtOS is a second brain for AI agents — a portable, self-owned memory layer that sits underneath all your AI tools. It connects via MCP, the universal protocol for AI tool interoperability, so every agent you use can read from and write to the same persistent knowledge base.

Unlike traditional second brain apps built for human note-taking, ThoughtOS is designed for cross-tool AI context sharing. Your memory should follow you to the tool you want to use — not the other way around.

How It Works

Capture once, recall everywhere

ThoughtOS processes every piece of knowledge through a five-stage pipeline that classifies, embeds, and connects your memories automatically.

01

Capture

Decisions, facts, insights, and preferences flow in naturally through conversation.

02

Classify

AI categorizes each memory: decision, insight, preference, person, or project.

03

Embed

Vector embeddings capture semantic meaning for intelligent retrieval.

04

Relate

Connections form between related memories, building a knowledge graph.

05

Recall

Semantic search finds the right memory at the right time, from any agent.

Core Value

Your second brain belongs to you

Not to one app, one platform, or one session. ThoughtOS is the memory operating system for your life and work.

01

Never repeat yourself

Your assistant retains important context across every session.

02

Personalized by default

Preferences and patterns compound over time.

03

Cross-project clarity

Ideas, people, and decisions stay connected.

04

Portable intelligence

Your memory belongs to you, not one app or session.

Features

A second brain that thinks

ThoughtOS isn’t just storage. It captures what matters, organizes it automatically, and makes it instantly retrievable through conversation. Think of it as a second brain that turns context into continuity.

01

Semantic Search

Find memories by meaning, not keywords. Powered by pgvector with HNSW indexes for blazing-fast similarity search across your entire knowledge base.

02

Health Scoring

Every memory has a health score based on freshness, usage, verification, dependencies, and neighborhood. Aging memories surface for review before they decay.

03

Pattern Synthesis

ThoughtOS doesn’t just store — it thinks. Automatic synthesis detects emerging patterns and surfaces insights you might have missed.

04

Cross-Agent Portable

Your memories aren’t locked to one AI. Claude connects via MCP, ChatGPT via REST API. Same brain, any agent, any device.

05

Lifecycle Management

Memories move through Active, Aging, and Retired states. Pin critical knowledge to keep it evergreen. Retire outdated facts automatically.

06

Topics & Spaces

Organize knowledge into topics and control visibility with spaces. Structure emerges naturally from your workflow.

Architecture

Built on open standards

ThoughtOS runs locally on your machine and connects to a cloud PostgreSQL brain. No vendor lock-in, no proprietary protocols.

AI Agents

Claude Desktop

35 MCP tools via stdio

ChatGPT

REST API via Vercel

Any MCP Client

Open protocol, open brain

Server

ThoughtOS MCP Server

Node.js / TypeScript
Runs locally on your machine
Classification & embedding pipeline

Cloud Brain

Neon PostgreSQL

Serverless Postgres + pgvector

Embeddings

text-embedding-3-small (1536d)

Claude LLM

Classification & synthesis

Who It's For

Anyone tired of losing context

Across chats, apps, and days.

// Agency Owners & Creators

Juggling clients, tools, and creative vision

You bounce between ChatGPT, Claude, and a dozen other tools daily. ThoughtOS makes sure every client preference, brand decision, and creative thread follows you everywhere — no re-explaining, no lost details.

// Corporate Professionals

Managing complexity across vendors, compliance, and deadlines

Events, projects, and relationships generate mountains of context. ThoughtOS catches conflicts before they become crises and surfaces the detail you captured three weeks ago — right when you need it.

// IT & Operations

One person holding together the entire infrastructure

When you’re the solo IT pro for 140 people, you can’t afford to forget a fix that worked two years ago. ThoughtOS turns your troubleshooting history into institutional memory that never leaves — even if you do.

// Teams & Organizations

Scalable second brains for the whole company

Deploy ThoughtOS across your team. Each person gets their own brain with their own context. Institutional knowledge lives in the system, not in someone’s head — onboard new hires in minutes, not months.

Demo

See the second brain in action

Capture a decision in one tool, then recall it from any other. This is what a session with ThoughtOS looks like.

Illustrative session
Economics

Costs almost nothing.

Your second brain runs on free tiers and open protocols. You choose every piece of the stack — from zero-cost to enterprise-grade. It’s your knowledge, built your way.

One example of many

$0.10
per month using free-tier services
ComponentMonthly Cost
Serverless Postgres (Neon/Supabase)$0.00
Embeddings (~20 thoughts/day)$0.05 - $0.15
LLM calls (classify + extract)$0.05 - $0.15
Total$0.10 - $0.30
Illustrative — actual usage varies.

Your stack. Your rules.

Every component is yours to choose. Start free with open-source tools, scale up with managed services, or bring enterprise infrastructure you already pay for. ThoughtOS is a specification — you decide what powers it.

Database
Free: Neon, Supabase, SQLite | Paid: Self-hosted Postgres, AWS RDS, PlanetScale
Embeddings
Free: Local models (Ollama) | Paid: OpenAI, Anthropic, Cohere
LLM
Free: Ollama, LM Studio | Paid: Claude, GPT-4, Gemini
Capture
Slack, Discord, CLI, Web UI, Email — use what you already have
MCP Server
Node.js, Python, Go — build in the language your team knows
The spec is the constant. The stack is the variable. Free, paid, or anything in between — you’re in control.
Next solution
Collections taking too much manual follow-up? — AR Collections
Build Your Second Brain

Ready to give your AI agents a second brain?

Whether you want early access, consulting, or a done-for-you deployment — we’d love to hear from you.

Our biggest bottleneck is .
type it in, then hit enter