Skip to content

Tagvico AI v2.0.0-alpha.1

Pre-release
Pre-release

Choose a tag to compare

@arturict arturict released this 12 Jul 10:04
adc0a88

Tagvico AI v2.0.0-alpha.1

This is the first public v2 prerelease. It is intended for representative
testing with an existing Paperless-ngx installation before stable 2.0.0.

Highlights

  • Review suggestions before writing them, or opt into Automatic mode per
    installation.
  • Inspect structured metadata diffs, apply or reject durable suggestions, and
    restore the first captured metadata snapshot.
  • Keep tag vocabulary controlled with groups, caps, and an exception queue.
  • Use local Ollama, direct APIs, compatible gateways, OpenCode Go, Ollama
    Cloud, GitHub Copilot, or experimental ChatGPT subscription access.
  • Recover OCR-poor documents, interrupted work, retries, and terminal failures
    from the Operations UI.
  • See token use, estimated model cost, history, runner state, and provider
    health from the web interface.
  • Optionally share coarse installation analytics. Sharing is off by default,
    uses rotating identifiers, exposes the exact payload locally, and requires an
    explicitly configured HTTPS receiver.

Upgrade from v1

  1. Stop Tagvico and back up the complete tagvico_ai_data volume.
  2. Replace deprecated pre-Tagvico environment variables with their documented
    TAGVICO_* equivalents.
  3. Pin ghcr.io/arturict/tagvico-ai:2.0.0-alpha.1; do not use latest for this
    prerelease.
  4. Start the container, inspect its logs, and verify /health and
    /api/health.
  5. Process synthetic or non-sensitive representative documents in Review
    first
    before enabling Automatic writes.

Do not run v1 and v2 against the same data volume simultaneously. Restore the
volume backup before rolling back if v2 has migrated its database.

Known prerelease boundaries

  • ChatGPT subscription access is experimental and is not an API SLA.
  • Model catalogs, entitlements, pricing, and quotas remain provider-controlled.
  • The REST/configuration/database contracts may still change before stable v2.
  • The reference analytics receiver must be deployed separately; no official
    collector endpoint is embedded in this prerelease.

Full documentation: https://tagvico.arturf.ch/docs/

Help validate v2

Test with synthetic documents, then share a redacted installation, upgrade, or workflow result. Reproducible bugs belong in the linked bug-report form; never include credentials or personal document data.