Document Management provides the full content lifecycle in one place: capture, storage, versioning, metadata and classification, full-text search, check-in and check-out, retention and access control. Documents enter by direct upload, by drag and drop, by inbound email capture, by scan, or automatically from any module in the suite as it generates output. On arrival each file is fingerprinted, deduplicated and written to durable object storage, with its business context, source, subject, author and classification recorded alongside it.

Every document is a living object rather than a static file. Users open, edit and return content through check-out and check-in, which locks the document against concurrent edits and records a new version on return, so the current copy is never ambiguous and superseded copies are never lost. Major and minor version numbering distinguishes substantive revisions from small corrections, and any prior version can be viewed, compared or restored. A rendition engine produces web-viewable previews and standardised PDF output so documents can be read in the browser without their native application. Folder and category hierarchies, saved views and virtual collections let teams organise the same underlying content in the way that suits their work.

The heart of the module is the distinction between a document as a business object and the versions that carry its actual content over time. A document is the stable identity: the thing people refer to, link to and hold rights over. Behind it sits an ordered history of versions, each an immutable snapshot of the file as it stood at one moment, with its own author, timestamp and integrity signature. The current version is simply the one presently in force; older ones remain permanently retrievable. This separation is what lets a link stay valid while the content beneath it legitimately evolves.

Around each document sits its classification: the descriptive and governing attributes that say what kind of thing it is, which part of the business it belongs to, how sensitive it is and how long it must be kept. Classification is not free-form tagging but a controlled vocabulary tied to document types, so a supplier contract, a payroll record and a quality certificate each inherit the handling their type demands. Finally, every document carries a relationship to the business records it evidences, the contract, invoice, ticket or asset it belongs to, so content and transaction are always two views of the same reality. Together these concepts let the module answer not only where a file lives, but what it is, what it proves, who may touch it and when it must go.

The principal workflows follow content from creation to disposal. A user or a module deposits a document, the module classifies it and applies the retention and access rules its type carries, and it becomes immediately searchable and linkable. When someone needs to revise it, they check it out, work on it, and check it back in as a new version with a note describing the change; colleagues see the lock while it is out and the fresh version when it returns. Where a document needs sign-off, it moves through a review and approval route, gathering comments and formal approvals before its status advances to released.

Retrieval is the workflow people run most. Full-text search reads inside the content itself, not just the metadata, and combines with faceted filters on type, owner, date, business context and classification to narrow millions of documents to the handful that matter. Saved searches, subscriptions and recent-activity views keep frequently needed content close. Document generation runs the reverse path: templates with merge fields draw live data from the suite to produce letters, contracts and statements as finished PDFs, filed straight back into the repository under the right classification. Legal hold suspends disposal on demand, and controlled export produces a complete, evidenced set for audit or disclosure.

The depth that matters in a repository is defensibility: proving that a document is authentic, complete and handled according to policy. Every version is written once and never altered, sealed with a cryptographic hash so any later tampering is detectable, and each is stamped with who created it and when. The full chain of custody, every view, edit, download, reclassification, permission change and deletion, is captured in an immutable audit trail that can be reconstructed for any document as it stood on any past date.

Retention is rule-driven rather than manual. Each document type carries a retention schedule that fixes how long content is kept and what happens at the end of its life, whether it is destroyed, archived or reviewed. Schedules trigger from meaningful events, the close of a contract or the end of a financial year, not merely from upload date, and disposition runs automatically with a certificate of destruction recorded for what is removed. Legal holds override any schedule and freeze affected documents until released. Access is governed at document and folder level in addition to module-level rights, so sensitivity classifications, need-to-know boundaries and segregation of duties are enforced on the individual file. These controls give the organisation a content estate that satisfies data-protection, financial and industry record-keeping obligations by construction, and stands up when it is examined.

Document Management is shared infrastructure, and nearly every module in the suite is a client of it. Contract Management stores executed agreements and their negotiated drafts here, keeping each clause version and signature against the contract record. Finance and Invoicing lodge supplier invoices, credit notes and remittance advices as the source documents behind every posting, so the ledger entry and its evidence are one click apart. Procurement files purchase orders and goods-received notes; Human Resources keeps contracts of employment, certifications and disciplinary records under the strict access and retention their sensitivity requires; Service Management attaches inspection reports, photographs and correspondence to tickets and assets.

Identity and access rights flow from the suite's central Access and Identity controls, so a person's permissions on content follow the same roles and boundaries they hold everywhere else. Because the repository is the single home for content, a document generated by one module, an invoice, a certificate, a signed order, is instantly available to any other module that legitimately needs it, without copies proliferating. Reporting and analytics draw on the repository to measure content volumes, retention exposure and outstanding approvals across the estate, giving the business one view of its documentary record rather than a dozen disconnected stores.

AI Workers act as first-class users of Document Management, holding their own identities, permissions and audit footprint exactly as human users do. They answer conversational questions against the repository: a user asks for the current signed version of a named supplier's master agreement, or every quality certificate expiring within ninety days, and the Worker resolves the query against content and classification and returns the documents with their context. They execute actions directly, filing, reclassifying, routing for approval, applying a legal hold or generating a document from a template, all within the permissions their role grants and all recorded in the same audit trail.

Workers read inside documents, not just around them. They extract structured data from uploaded content, pulling parties, dates, amounts and terms from an incoming contract or the totals from a scanned invoice, and write it back as metadata or hand it to the owning module. They monitor the estate for anomalies and exceptions, a document checked out and never returned, a retention date reached without disposition, a classification inconsistent with content, a sensitive file exposed too broadly, and raise these for attention before they become failures. In review and approval routes a Worker can serve as a formal node, checking a generated document against policy, verifying that a required signature or clause is present, and either approving, escalating or returning it with its reasoning attached. Throughout, the Worker offers decision support, surfacing prior versions, related documents and relevant precedent so the person deciding sees the full picture, while every judgement it makes remains explainable and on the record.