Tagvico AI v2AI filing for Paperless-ngx, under your control

Self-hosted, reviewable AI metadata for Paperless-ngx with your choice of local or hosted model.

Install v2

Explore features

Choose a provider

🗂️

Structured filing

Generate titles, tags, correspondents, document types, dates, languages, custom fields, and optional owner assignments.

✅

Review first or automate

Queue suggestions for approval or write validated metadata directly. Change modes without losing queued work.

🔌

Your model, your boundary

Use local Ollama, direct APIs, OpenRouter, Azure, compatible gateways, GitHub Copilot, or experimental ChatGPT subscription access.

📈

Visible operations

Inspect progress, processing history, usage, retries, OCR recovery, rescan actions, and metadata restoration from the web UI.

💬

Optional Telegram access

Give allowlisted family members natural-language search, cited downloads, follow-ups, and uploads through their own Paperless tokens.

Start here

Tagvico AI is a self-hosted companion for an existing Paperless-ngx instance. It reads OCR text already stored in Paperless, asks your chosen model for a structured filing suggestion, validates that suggestion, and either queues it for review or writes it back.

The v2 documentation is frozen under this URL when a future major version is published. Use the version menu in the navigation to move between major versions without losing the instructions that match your installation.

Stable v2

Version 2.0.1 is the current stable v2 release. Pin the immutable image version, back up the data volume before upgrades, and start new installations in Review first mode.