Skip to main content

Enterprise rollout roadmap

This page describes a practical rollout path for a serious customer pilot.

Phase 1: Discovery​

Collect:

  • document types;
  • volume per month;
  • sample files;
  • current process;
  • ERP system;
  • approval rules;
  • source channels;
  • language and country requirements;
  • security requirements;
  • archive and retention requirements.

Output:

  • product profile selection;
  • pilot scope;
  • success metrics;
  • integration list.

Phase 2: Tenant setup​

Configure:

  • tenant or customer group;
  • companies;
  • clients;
  • roles;
  • product scope;
  • inboxes;
  • OCR/AI provider profile;
  • ERP profile;
  • archive retention.

Output:

  • working sandbox;
  • seed users;
  • first intake route.

Phase 3: Golden dataset​

Build a verified dataset:

  • 100-200 documents for pilot;
  • expected header fields;
  • expected line items;
  • supplier list;
  • known exceptions;
  • anonymization if needed.

Output:

  • baseline quality score;
  • provider comparison;
  • expected automation percentage.

Phase 4: Workflow pilot​

Run real scenarios:

  • upload batch;
  • email or Drive import;
  • OCR/AI;
  • verification;
  • approval;
  • ERP dry-run;
  • archive.

Output:

  • pilot report;
  • issue list;
  • provider routing decision;
  • field processing rules.

Phase 5: Production hardening​

Complete:

  • real connector credentials;
  • connector health dashboard;
  • SSO and 2FA;
  • production storage;
  • backup and restore;
  • monitoring;
  • audit review;
  • API rate limits;
  • worker queue retry and dead-letter handling.

Output:

  • production checklist;
  • go-live decision;
  • support runbook.

Phase 6: Go-live​

Start with limited scope:

  • one company or client group;
  • one document type;
  • one ERP profile;
  • one intake route;
  • daily monitoring.

Then expand:

  • more clients;
  • more sources;
  • more document types;
  • archive retention;
  • public API;
  • mobile app.

Success metrics​

  • percentage of documents processed without human correction;
  • average verification time;
  • export success rate;
  • approval SLA;
  • duplicate detection rate;
  • field-level accuracy;
  • line item accuracy;
  • cost per document;
  • customer satisfaction.