Cookies
We’d like to use analytics and advertising cookies to understand how the site is used and which ads bring visitors here. They stay off unless you allow them. See our privacy policy.
QDocs · Intelligent document processing
QDocs turns incoming packages into individual, classified documents on the right record, and turns what’s inside them into data your workflows act on, with a confidence on every value and a person reviewing anything uncertain.
Split & classified
| First notice of loss | p. 1–3 |
| Police report | p. 4–9 |
| Repair estimate | p. 10–14 |
| Photos | p. 15–41 |
Extracted · claim fields
| Field | Value | Confidence |
|---|---|---|
| Date of loss | 2026-08-14 | 0.97 |
| Estimate total | $8,412.50 | 0.94 |
| Policy number | HO-48213 | 0.41 Review |
Orchestrated, not bolted on
Most document AI stops at output: a file of extracted fields that someone still has to load, check, and act on. In QDocs, a workflow step calls the model and the answer comes back as ordinary data on the record, where your rules, worklists, and next steps are already waiting.
Why QDocs
Mixed packages are split into individual documents, each one is classified against the document types you define, and every document lands on the record it belongs to. No manual prep, no re-indexing.
The fields you care about are extracted as typed values (dates, amounts, yes/no, text), each with its own confidence. Rules, workflows, and reports use them like any other data.
Confident results flow straight on. Uncertain ones go to a person, on a worklist, with the source document beside the value. Every step runs inside your security and your audit trail.
How it works
Each stage is a step in a QBO workflow, so you choose which stages run, in what order, and for which document types. Stages marked AI · built in call the model you configured; the rest are standard platform operations.
Upload, drag and drop, the API, inbound email, or a watched sFTP or cloud folder.
Lands asA document on a record, in your storage
A multi-document PDF is broken into its individual documents.
Lands asOne document per segment, linked to the source package
OCR extracts the text, so every page is searchable.
Lands asSearchable text on the document
Each document is matched to one of your document types, and labeled.
Lands asThe document’s template and labels
The fields you define are read from the text, or from the original PDF or image.
Lands asA score: typed values, each with a confidence
Rules send low-confidence or missing values to a worklist for a person to check.
Lands asA task on the right worklist
Workflows, rules, and decisions use the data: open a case, generate a letter, notify a partner.
Lands asThe next step in your process
Classification can also run in the same pass as splitting, with the fields for each document type extracted at once. See how QBO manages documents and how scores store extracted values.
Capabilities
Document types, labels, and fields are templates. Describe each one the way you’d brief a new colleague, and QDocs uses that description as its instructions. Adding a document type is a template edit, not a project.
Send a compound PDF to a multimodal model and get back one document per segment, each typed and linked to the original package. Or split by bookmarks, explicit page ranges, or file size, with no AI at all.
The model chooses among your document types using the plain-English description on each one, running OCR first when it needs to. Unmatched documents fall back to a default type you choose.
Apply every label that fits, such as language, sentiment, or a review flag, from labels you describe. Existing labels aren’t duplicated.
Write a prompt for each field, choose its type, and optionally weight it. Every value comes back with a confidence, and the document gets a weighted confidence across all fields.
When layout, tables, handwriting, or images matter, the model reads the original PDF or image instead of OCR text.
Turn a document into new records through the standard import path, including nested children, such as line items under an invoice, with a preview before anything is saved.
Transcribe recorded calls and voicemail (mp3) into searchable text, and optionally score the transcript in the same step.
Find identical files by hash before they’re processed twice, and file bulk-loaded documents to the right type and record from their file names.
Use cases
A loan file, a claim packet, a lease, an application: the documents differ, the job doesn’t. Here’s how teams in each industry use QDocs.
A loan file arrives as one scan or a stack of uploads. QDocs turns it into an indexed, verified file.
A claim packet mixes forms, reports, estimates, and photos. QDocs sorts it before an adjuster opens it.
Leases and property files are long, inconsistent, and full of dates that matter.
Credit unions, lenders, and insurers all onboard people with documents.
Study and site documentation is high volume, regulated, and audited.
See how QDocs fits into mortgage servicing, insurance, and credit union operations.
Scale & cost control
The same configuration runs a single document in real time or a backlog as a provider batch job. Token use is managed and accounted for, so AI cost is a line item, not a surprise.
Integration
Documents stay in your storage: Amazon S3, Azure, Google Cloud, sFTP, or a network share. Every QDocs stage is an operation on the same API pattern as the rest of the platform, and long-running work is queued for app servers so your intake never waits.
Split a package (queued)
GET /api/attachment/queue/pdfsplit?ID=2207
Classify a document
GET /api/attachment/aiclassify?ID=2208&AgentName=Gemini
Label a document
GET /api/attachment/ailabel?ID=2208&AgentName=Gemini
Extract fields into a score (queued)
GET /api/score/queue/calculate?Template=Repair Estimate&Object=Attachment&ObjectID=2208
Works with
FAQ
QDocs is Quandis’ intelligent document processing product. It splits document packages, classifies each document, extracts the data you need with a confidence for every value, and routes uncertain results to people for review. Because it runs on the QBO Platform, the results drive your workflows, rules, and worklists directly.
No. You describe each document type in plain English and write a prompt for each field you want extracted. Changing behavior is a template edit, not a model training project or a code change.
QDocs works with Anthropic Claude, Google Gemini, OpenAI, and Amazon Bedrock. You choose the provider and model for each task by configuration and can change them without code.
Every extracted value carries its own confidence, and each document gets a weighted confidence across its fields. Rules use those values like any other data, so you decide which results flow straight through and which go to a worklist for a person to check.
Yes. A multimodal model can identify where each document starts and ends, and QDocs saves each segment as its own document, typed and linked to the original package. PDFs can also be split by bookmarks, explicit page ranges, or file size.
Documents stay in the storage you configure, such as Amazon S3, Azure, Google Cloud, or sFTP. AI methods run as the calling user, on records that user is allowed to reach, and send content only to the provider you configured. Results come back as ordinary data inside your audit trail. Security documentation is available in our Trust Center.
Send large volumes as provider batch jobs through QBO’s queue, cache long instructions where the provider supports it (Gemini), and post token usage to a ledger at per-model rates so cost is visible per record, client, or month.
QDocs trials run for 30 days on your own sample documents. We’ll agree on the document types, fields, and success criteria with you first. Request a demo to get started.
Documentation
Set up splitting, classification, and extraction, and call them from the API.
Read the QDocs guide →Built on QBOQDocs runs on the same orchestration platform as everything Quandis builds — cases, documents, rules, audit trails, and partner-safe security included.
Explore the QBO Platform →Your privacy matters to us. We never sell your personal information to anyone.
Booked
We're looking forward to meeting with you. Your calendar invitation is on its way: please check your email for the confirmation and meeting details.
See how the QBO platform works →Browse product documentation in Learn →