Skip to main content

ARCIS by Fingress · document solutions

ARCIS Document Intelligence & Governance

A governed repository of record for every document, with AI extraction and maker-checker control at every step.

Overview

Why ARCIS

Regulated businesses re-key the same documents, re-check the same evidence and struggle to show an auditor what happened. ARCIS, the document solutions family born from Fingress, stores every document once in an enterprise content repository built on Apache Oak, reads it with AI, routes it through dual-control approval and keeps a complete, defensible trail from upload to disposal, on the storage your organisation already trusts.

Who uses it

  • Makers, checkers and approvers in operations
  • Back office, legal and compliance teams
  • Records managers and auditors
  • IT and enterprise architecture teams

Capabilities

What ARCIS Document Intelligence & Governance includes

Governed document workflow

Every document moves through a content-type specific approval chain: draft, submitted, review, compliance review, approved or rejected, archived.

AI extraction and classification

Classify documents, extract fields with confidence scores, and route low-confidence results to a manual review queue instead of letting them through.

Records management

Retention schedules, legal hold, scheduled disposition review and authorised destruction.

Search

Free-text and faceted search across names, content, tags, content types and workflow states.

Versioning and check-out

Full version history and exclusive edit locks so concurrent editors never overwrite each other.

Templates and document generation

Vetted templates, typed or Word-based, filled with transaction data to issue branded documents and PDFs.

Functional

Everything a regulated document lifecycle needs

One consistent functional pattern for any regulated industry: a maker authors or uploads, a checker reviews, back office scrutinises, legal vets non-standard documents, an approver signs off, and an auditor traces it all.

  • Content types and rules

    Define content types with metadata schemas, allowed file extensions, size limits, renditions and their own approval chain. Tenants author their own, or start from a curated catalogue.

  • Maker-checker at every step

    Segregation of duties is enforced, not just recorded: the person who submits a document cannot approve it, and every return or rejection requires a remark.

  • Permanent documents

    Upload once, verify once. KYC and customer documents verified by back office are reused across later transactions without re-scrutiny.

  • Confidence-based review

    Extraction results are scored. High confidence is accepted automatically, borderline results are flagged for spot checks, and low confidence goes to a manual review queue.

  • Records, retention and legal hold

    Retention policies per content type, legal hold that overrides disposal, a scheduled disposition review that flags (never deletes) expired records, and destruction only on explicit authorisation.

  • Collaboration and file store

    Folders, sharing with users and groups, annotations pinned to a position on a document, pins, notifications, and a versioned general-purpose file store.

  • Audit observations and dashboards

    Auditors raise, track and close observations against records, with dashboards across tenants and content types.

  • Multi-tenant administration

    Tenants, users, groups and custom roles managed per tenant, with per-tenant branding and storage quotas.

Integration

Fits into the systems you already run

ARCIS is built to be called by your channels and core systems, and to bring legacy content in without a big-bang cutover.

  • REST API

    Every capability, from upload and approval to search and records actions, is available through a tenant-scoped REST API with consistent error responses.

  • Java client SDK

    A ready-made Java client with retries, circuit breaker, token handling and metrics, also shipped as a single shaded JAR for application servers such as WebSphere.

  • Core system integration

    Push and pull transaction data to and from core systems, such as trade finance back offices, as documents move through their lifecycle.

  • Bulk migration of legacy content

    Resumable, checksum-verified migration from file systems, Amazon S3, Google Drive, JDBC databases and HTTP archives, with durable orchestration that survives restarts.

  • Pluggable AI providers

    Extraction runs through a provider interface. Google Cloud Document AI is supported today, with an offline provider for testing and room to add others.

  • Rendering and text extraction

    Apache Tika extracts text for search, and LibreOffice renders Office documents for preview and PDF output.

Security & compliance

Controls designed in, not bolted on

Security works at four independent layers: who you are, what you may do, whose data you can reach, and a tamper-evident record of what you did.

  • Authentication

    Stateless JWT bearer tokens and HTTP Basic, with bcrypt-hashed passwords. Tokens verify on any node, so clusters need no shared session store.

  • Capability-based authorisation

    Permissions such as read, write, review, approve and archive are composed into each tenant’s own roles, and checked on every service call.

  • Deny-by-default access control

    Documents and folders are private until shared. Access is granted directly or to groups, and cascades through folders.

  • Fail-closed tenant isolation

    Each tenant’s content lives in its own repository subtree, and every request is checked against the caller’s home tenant. Mismatches are refused.

  • Governance floor

    The platform sets a minimum control level per tenant. Tenants can design stricter approval chains, never weaker ones, so four-eyes cannot be designed away.

  • Dual control on administration

    Creating, disabling or enabling users requires a second administrator’s approval.

  • Append-only audit trail

    Tenant, content-type, user and document events are recorded in an append-only log, with remark and version history on every record.

  • Protected credentials

    Credentials for migration sources are encrypted at rest.

Built on Apache Oak

An open content repository standard, trusted by enterprises

ARCIS is built on Apache Jackrabbit Oak, the open-source, standards-based (JCR) content repository from the Apache Software Foundation, used at scale in enterprise content and digital experience platforms. Fingress chose Oak as the framework for ARCIS and builds document governance on top of it, so your records sit on a proven, vendor-neutral foundation rather than a proprietary store.

  • Standards-based: Java Content Repository (JCR) API, hierarchical content with versioning
  • Open source under the Apache License 2.0, with no repository lock-in
  • Scales from a single node to a cluster sharing the same content store
  • Storage is a configuration choice: the application does not change when the back-end does
  • Separate control plane for users, permissions and audit on PostgreSQL or Oracle
ARCIS architecture: document governance services run on the Apache Oak content repository, which persists content to a pluggable node store and binaries to a pluggable blob store.

ARCIS by Fingress

Document governance

Maker-checker workflowAI extractionRecords & retentionAccess & audit

Content repository

Apache Jackrabbit Oak · JCR API

Versioning · hierarchical content · indexing · clustering

Content (node store)

  • Segment files
  • PostgreSQL / Oracle
  • MongoDB

Binaries (blob store)

  • File system / NFS
  • PostgreSQL / Oracle
  • Amazon S3 / MinIO

Storage

Keep documents where your organisation wants them

Choose the storage back-end per deployment. Changing it is configuration, not a re-build, and the governance layer works the same on all of them.

File system or NFS

Binaries in a de-duplicated file data store on local disk or a shared NFS mount.

Oak FileDataStore. Simplest single-node option uses Oak segment storage with no external dependency.

Relational database

Content and binaries in the same PostgreSQL or Oracle database you already operate and back up.

Oak RDB document store and RDB blob store. Several application nodes form one cluster.

Amazon S3

Binaries in Amazon S3 buckets or S3-compatible object storage such as MinIO.

Oak S3DataStore, combined with any content store.

MongoDB cluster

Content tree on MongoDB for horizontally scaled, multi-node deployments.

Oak Mongo document store, with binaries on any of the options above.

Powered by the Fingress platform

Configure it to fit, extend it when you need to

Every Fingress solution runs on the same digital platform and building blocks, so you are never limited to what ships out of the box.

See ARCIS Document Intelligence & Governance in action

Book a walkthrough tailored to your products, customer segments and systems.