Hands on · Classify
Check a document at the door. Seal it for the record.
Redlineo's Classify works on one document — no pair, no redline. Drop an incoming file and it is checked against your conventions, scrubbed by your policy, and sealed with a tamper-evident record — with a runtime proof that the content itself was never touched. The document intake check your repository has been missing.
One drop, four things
Extract. Validate. Seal. Prove.
-
1 · Extract + check
Document metadata extraction & validation
Your configured extraction runs — company, state, form number, whatever your conventions declare — and your assertions are evaluated against the single document. Checks that need two documents are skipped and explicitly recorded as skipped, never silently dropped.
-
2 · Sanitize
Your scrub policy, applied
The deployment's metadata-scrub policy applies exactly as it does on every produced output — author names, custom properties, whatever the configuration declares.
-
3 · Seal
A tamper-evident document seal
The classification record — extracted values and assertion verdicts — is embedded in the file, then the integrity manifest is written as the very last step. The sealed copy verifies on the Verify page like every other Redlineo output.
-
4 · Prove
Content untouched, proven at runtime
The document's text content is fingerprinted before and after; the two must be identical. That is a per-request proof — shown as the ✔ badge — that classification changed properties, never content.
Verify document integrity — anytime
The seal is checkable years later.
Drop any sealed document on the Verify page: every part of the file is re-hashed from the uploaded bytes and checked against the embedded manifest. No database, no lookup — the file proves itself.
- ✔Integrity.
ok— unaltered since sealing;failed— something was modified, and the failing parts are named. - ✔Origin. With a signing secret configured, the manifest carries an HMAC only your servers can produce — a look-alike file is exposed.
- ✔The record. The embedded classification — what was found, what failed, what was skipped — travels inside the file, under the seal.
Where it earns its keep
Two jobs, one primitive.
-
Intake gate
Check before it lands in the repository
Point your ingest flow at Classify: every incoming filing is checked against naming and metadata conventions before it is stored, and what you store is the sealed, re-verifiable copy. One REST call per document — a one-step quality gate with a durable receipt.
-
Retroactive sealing
Give existing documents the same proof
Run documents already in your repository through Classify to give them the same integrity and metadata record your comparison outputs carry — so the whole store, old and new, answers an audit the same way.
In the UI it is one page (/classify) with the same caller-selected profiles as Compare; in a pipeline it is a single call. The Classify page, step by step → · The API at a glance →
Part of the same proof stack
The ingest-side sibling of a verified comparison.
The same evidence Redlineo attaches to every comparison output — the embedded record, the integrity manifest, the one-drop verification — applied to single documents at the door. Self-hosted, deterministic, nothing stored.