Molly Reader

The old paper is still there; the data isn't

Most organizations have digitized a great deal already, yet cabinets of older paper records remain untouched — most visibly the files of elderly residents at ward and commune offices. The familiar result: pulling a single record takes five to fifteen minutes, service slows down, and people end up re-submitting paperwork the office should already have.

Molly handles exactly that backlog. You define a template once, feed in a whole batch of scanned documents, and get back structured data you can search in a second instead of digging through folders. The part that matters most: everything runs on your own infrastructure, and your data goes nowhere.

What Molly actually does

Molly digitizes paper documents and extracts structured data. It is not a chatbot you talk to about your files. The job is concrete: turn scanned documents into clean fields — names, dates, reference numbers — that you can search and push into other systems to keep working with.

It fits ward and commune offices that need to finish digitizing civil-status records, and any organization sitting on contracts, forms, receipts, or HR files that should become usable data.

How it works

1. Build a template, no code required. Declare the fields to capture for each document type and flag the sensitive ones that hold personal data. An AI suggestion drafts the template from a sample document so you're not starting from a blank page.
2. Upload in bulk. Drop in scanned images or PDFs; multiple pages merge into one record, with duplicates caught automatically.
3. AI reads and extracts. The vision model classifies each document, then extracts against your template. It does the reading itself — there's no separate OCR engine to bolt on.
4. A person reviews. You check the original side by side with the pre-filled form, fix anything that needs it, and approve. Every action is logged.
5. Store and use. Approved data lands in an append-only database with instant full-text search, ready to push into other software through an API or webhook.

Your data stays on your infrastructure

Molly runs Google's open Gemma vision model fully on-premise, via Ollama or LM Studio. Not a single line of text leaves for the internet or any third party, which is why Molly is safe even for confidential records containing personal information.

That's also what makes data sovereignty real here: on-prem processing, no egress, sensitive-field marking, a complete audit trail, plus login and role-based access. Molly is designed to support compliance with Vietnam's Personal Data Protection Law and in-country data residency requirements. And unlike a heavy enterprise rollout, getting Molly running on-prem is light and quick rather than a months-long project.

Transparent pricing

• License: free.
• Standard installation: 20,000,000đ.
• Hardware: you provide it, sized to your volume and the throughput you want — we'll advise on the right spec for your document load.
• Custom development: billed by man-day, on request.

See it on your own documents

Don't take a marketing number on faith — put your own documents into the public demo and watch how accurately the AI reads them. Anything you test auto-deletes at the end of the day. For hardware sizing and an on-prem rollout, reach us at contact@tetra.vn.