Skip to content
UsecaseBoard
← All use cases
githubLocal AI

Query household invoices through a local document stack

At a glance

A home-lab author connected Paperless-ngx to local AI while keeping invoice fields and calculations tied to its API. This workflow is listed under Local AI, Document Management, Household Paperwork.

SourceBuilding a tool for these workflows? See weekly rates.

What they did

1) Start with a deployed copy of fwinchi/ia-local-casa: the author's Windows/Docker Desktop setup with Paperless-ngx, AIssist, Ollama and Open WebUI. This workflow requires that custom stack, storage access and a Paperless API token. Put your token in scripts/secrets.local.bat using its example file. 2) Create gptoss-paperless from ollama/gptoss-paperless.Modelfile and use the supplied prompts/openwebui-gptoss.md instructions in Open WebUI. Start scripts/start-mcpo.bat and connect its localhost:8001 tool endpoint. Those instructions route invoice fields and calculations through Paperless tools rather than semantic retrieval or model arithmetic. 3) Copy an invoice into Paperless's consume folder. Wait for OCR and AIssist field extraction, then check the archived document in Paperless. 4) In Open WebUI, select gptoss-paperless and enable only the Paperless tool. Ask about an invoice or a filtered total. 5) Open the returned document and verify the answer against the original invoice.

How it went

The author reported that enforcing API retrieval for structured invoice fields fixed most previously incorrect answers.

Worth knowing

This is a custom integration. A discussion reply identifies stale-index risks when source files change without reindexing.

Save ideas in this browser. Upvotes count across the board.

Prompt excerpt

No prompt excerpt was provided by the source.

Sponsored

CodeBuddy

Tencent Cloud AI coding assistant for IDE and CLI.

Try CodeBuddy ↗
Explore all →