TAILOR — Persistent AI Memory. 100% Local. Open Source.

The Problem

AI assistants lose context when tabs close.

Every new session is a blank slate. Your decisions, preferences, and history vanish — requiring constant re-explanation.

The Solution

T[AI]LOR adds a persistent memory layer to your hardware.

A dedicated local service that enables any LLM to recall years of conversations, preferences, and facts — instantly.

01 — PRIVACY FIRST

100% Local

Your memories never leave your machine. Zero cloud dependency, zero data leaks.

02 — COMPATIBILITY

LLM-Agnostic

Switch between Claude, GPT, Gemini, and Ollama without losing a single byte of context.

03 — PROTOCOL

MCP Native

20+ built-in tools via Model Context Protocol. Works with Claude, ChatGPT, and any MCP client.

The Senses — Multi-Source Ingestion

folder_open

Files

Recursive scans of PDF, Excel, CSV, Word, and PowerPoint.

alternate_email

Communications

Gmail (OAuth2), IMAP sync, and Telegram real-time capture.

cloud_sync

Cloud Storage

OneDrive, Google Drive, Dropbox via rclone.

history_edu

AI History

Parsers for ChatGPT, Claude, and Gemini Takeout exports.

upload_file

Browser

Drag-and-drop dashboard uploads with progress tracking.

System Architecture

description Local Files

mail Email Streams

cloud Cloud Storage

history AI History

upload Browser Upload

CORE ENGINE

T[AI]LOR

Enrichment · Search · Memory

terminal MCP Tools

send Telegram Bot

dashboard Web Dashboard

Feature Stack

Enterprise-grade primitives for local AI

01

Hybrid Search

Semantic + keyword + entity search with ONNX cross-encoder reranking on consumer hardware.

~28ms / query

02

Atomic Facts

LLMs extract discrete facts from every document. Second-order derivation infers connections you never made explicit.

03

Temporal Grounding

Dual-layer dating — document_date vs event_date — for time-aware retrieval. Memory isn't just what, it's when.

04

Nightly Enrichment

Automated pipeline: entity extraction, fact derivation, supersession filtering — all while you sleep.

05

20+ MCP Tools

Built-in autonomous tools for search, memory management, reminders, Telegram, file operations, and system commands.

06

Fact Supersession

When facts change, TAILOR detects contradictions and auto-supersedes outdated info. No manual cleanup.

07

Model Advisor

Scans HuggingFace and cloud APIs for models that fit your hardware. One-click install from the dashboard.

08

Multi-Backend Enrichment

Ordered list of LLM backends with automatic rotation on rate limits. Config-driven, zero downtime.

Manage
everywhere.

Whether you prefer a graphical dashboard or a conversational bot, TAILOR provides unified control over your local intelligence.

dashboard

Web Dashboard

Real-time KB stats, live config editor, 10-step setup wizard, conversation uploads.

send

Telegram Bot

Query your KB from your phone. Auto session capture after 10 min of inactivity.

smart_toy

TAILOR BOT

When did I last discuss the architecture refactor?

Tuesday at 2:15 PM during the sync. You decided to move to ONNX reranker...

Memory Benchmarks

Feature	T[AI]LOR	Supermemory
Hosting	100% Local / Self-Hosted	Cloud Managed
Privacy	Data on your machine	SaaS Terms
LLM Support	Any provider (4 built-in)	Locked to their stack
Tool Use	20+ autonomous MCP tools	API Webhooks
Cost	Free (Apache 2.0)	Monthly Credits
Model Discovery	Hardware-aware recommendations	—
Enrichment Backends	Multi-backend with auto-rotation	Single provider

Get Running in 60s

From zero to persistent intelligence

1

Install

git clone https://github.com/tailormemory/tailor
cd tailor && ./setup.sh

2

Configure

Open localhost:8787 — the 10-step Setup Wizard configures providers, sync, and Telegram.

3

Connect

Add TAILOR as an MCP connector in Claude, ChatGPT, or any MCP-compatible client.

AI Assistants Forget. T[AI]LOR Fixes That.