TensorFrog.it
PLEN

Capabilities · Data & reports

Document management — register, workflow and archive in the company process

The scope is a document system that operations, accounting and customer service actually open: a register, intake, classification, versions, approval workflow and search. The file sits in storage. Metadata, rights and history sit in the application. Start with one document type — not every file in the company at once. TensorFrog delivers software in an agreed scope. It is not a state archive, a notary or a KSeF tax adviser.

Documents

A register instead of a file folder

A document in the system has a register number, a type, an owner, a status and a current version. A network share or a mailbox does not give that picture: the same file comes back under another name and nobody knows which copy is in force. The register is where the document exists as a process record — not as a path on a disk. Preview and download apply to a named version. An edit lock when someone is already on the file. After intake a trail remains: who, when, from which source.

How we will judge the effect

We agree the goal and how we will measure it before work starts. We do not promise percentage savings without a baseline.

  • Time to find a document — from the query to opening the right version in the register.
  • Share of documents with required metadata, status and owner in the agreed type.
  • Number of copies outside the system — drives, inboxes, chats — in the agreed area.

Document intake — scan, upload, mail

A document enters through the UI, an API, a scan or a company mailbox — when mail is in scope. The system assigns a register number and a preliminary type. An attachment does not stay in a year-old thread: after classification it lands in the register or attaches to a case that already exists. Scan and upload require the same fields as intake through the API. The intake source stays in the history. A bulk drop of a folder without a type map is not the first stage.

Classification, types and metadata

Each type has its own fields: counterparty, case, order, period, status, tags from a dictionary. Required fields on intake — no drop-and-forget. A named person in the company keeps the tag dictionary. Typical first types: contract, protocol, cost invoice, case correspondence. Classification by hand or with a rule hint; a hard automaton only where an error must not pass. Further types — a later stage.

Versions, statuses and approval workflow

A new version does not silently overwrite the previous one. The register shows the current version and the history. Status — draft, pending approval, in force, archived — is a process field, not a file name. The flow: who submits, who approves, who returns with a comment. A rejection goes back to the owner with a reason. After approval the document is the in-force version in the register. An electronic signature — only as an integration with a service the company already has or names; TensorFrog is not a signature issuer.

Audit trail and access rights

Roles: who intakes, who approves, who only reads, who exports. A document a role cannot see does not appear in search. A change to metadata, version, status or rights has a trail: who, when, from which source. Export and download too. Log retention follows an agreed policy. This does not replace a state archive or a notarial procedure — it is a trail in software the company maintains.

Search, registers and retention

Two layers: full text (PDF and Office content where it can be indexed) and filters on metadata. A query joins a phrase with type, counterparty, case, period and tag. The result shows version and right. Folders and registers arrange the operational view — they do not replace metadata. Retention and archive follow an agreed policy — not everything forever. A recycle bin with a restore window. Migrating old folders is a separate scope: a map first, then moves in batches.

Tied to the case and integrations

A document belongs to a case, an order, a counterparty or a ledger record — when that link is in scope. Opening the case shows the documents; opening the document shows the case. The store is the one that already stands: a file share, object storage (S3 or equivalent), Azure Blob or an existing repository. The application does not keep binaries in the database. The database — PostgreSQL, MSSQL or MariaDB, per the company stack — keeps metadata, path and rights. CRM, ERP, mail, tickets and KSeF enter when the integration is in the offer. A KSeF invoice can land in the register and attach to a case; TensorFrog does not keep the books and does not replace a tax adviser. Each system has its own key, limit and log.

First stage

One document type, one store, required metadata, two roles and search on those fields. Preview of the right version. After sign-off: further types, an approval workflow, a content index, an inbound mailbox, a link to a case or an archive migration — each with its own quote.

Related areas

SQL & dataAPIKSeFOperational platforms
Data & reports

Describe your business need — we will propose the right technology and the next work stages. Contact: [email protected]

Go to contactView all capabilities