Robotic Process Automation (RPA)

The Automation Solution Delivery Centre (ASD) helps UBC units automate routine, rule-based, repetitive business processes using Robotic Process Automation (RPA). Software "bots" perform high-volume digital tasks — such as data entry, reconciliation, and report generation — the same way a person would, but faster and consistently.

We build on enterprise-grade automation platforms licensed and supported by UBC IT.

What RPA does well

RPA is a strong fit for processes that are:

• Rule-based, with clear decision logic and few exceptions
• Repetitive and high-volume (worth automating)
• Stable — the underlying systems and steps don't change often
• Based on structured, digital data (forms, spreadsheets, system records)

Examples of processes that fit well at UBC:

• Entering supplier invoices and expense data into Workday Finance from standardized submissions
• Bulk updates to employee or student records each term (appointments, TA/RA setup, award disbursement checks)
• Moving data between Workday and faculty- or unit-specific systems that don't integrate natively
• Generating and distributing recurring reports (enrolment, financial, HR) on a schedule
• Routing and acknowledging high-volume form submissions (onboarding checklists, access requests)

What RPA does not do well

To set expectations honestly, processes like these are poor candidates:

• Judgment calls. Admissions decisions, grade appeals, academic advising, or scholarship adjudication — bots follow rules and can't make discretionary decisions.
• Unstable processes. Workflows that change every term or depend on systems undergoing frequent updates; bots break and need rework.
• Low-volume tasks. If a task takes minutes a week — say, a form processed a few times a month — automation rarely pays back the build and maintenance effort.
• Fixing a broken process. Automating an inefficient process automates the inefficiency. We often recommend process improvement first.

Where AI extends RPA

Combining RPA with AI opens the door to automating processes that were previously out of reach:

• Document understanding — AI can extract data from semi-structured documents — scanned invoices, transcripts, application forms — that classic RPA can't read, routing low-confidence results to a person for review.
• Classification and triage — AI models can categorize incoming requests or emails (e.g., sorting helpdesk or program-inquiry emails) before a bot routes them.
• Generative AI assistance built into our platforms speeds up building and maintaining automations.

AI-assisted steps introduce probabilistic results, so we design these with human-in-the-loop review and follow UBC's privacy and information-security requirements for any personal or sensitive data.