USA
Canada
UK
Europe
Australia
Singapore
Japan
IndiaQ1. What is RPA and what processes are good candidates for it?
Short answer: Software bots automating rule-based, repetitive, high-volume digital tasks.
Strong answer: RPA uses software robots to mimic human actions across applications for rule-based, repetitive, high-volume, stable processes with structured input (data entry, reconciliation, report generation). Good candidates are well-defined and standardized; poor candidates are highly variable, judgment-heavy or frequently-changing processes.
Likely follow-up: What makes a process a poor RPA candidate?
What the interviewer is checking: Automation suitability judgment.
Common mistake: Automating unstable or judgment-heavy processes.
Q2. How do you approach exception handling and resilience in RPA?
Short answer: Categorize business vs system exceptions, retry appropriately, log and alert.
Strong answer: Design bots to distinguish business exceptions (data/rule violations — log and skip) from system exceptions (app/environment issues — retry, then escalate), with try-catch, retry logic, screenshots and detailed logging. Resilient bots recover gracefully and don’t lose transactions — often via a queue-based transactional model.
Likely follow-up: How do queues improve bot resilience?
What the interviewer is checking: Reliable automation design.
Common mistake: No exception strategy, so one failure stops everything.
Q3. What is the difference between attended and unattended automation?
Short answer: Attended runs with a human on their machine; unattended runs autonomously, usually scheduled.
Strong answer: Attended bots assist a human at their desktop, triggered by the user for front-office tasks. Unattended bots run without human interaction, typically scheduled/triggered on servers for back-office, high-volume processing, managed centrally. Many solutions combine both. The choice shapes infrastructure and licensing.
Likely follow-up: Which suits high-volume overnight batch processing?
What the interviewer is checking: Deployment-model understanding.
Common mistake: Confusing the two models and their use cases.
Q4. How do you calculate ROI and prioritize RPA opportunities?
Short answer: Weigh volume, time saved, error reduction and stability against build/maintenance cost.
Strong answer: Estimate benefit (hours saved × volume, error/quality gains, compliance) versus cost (development, licensing, maintenance) and feasibility (process stability, structured data). Prioritize high-volume, stable, high-effort processes. A process that changes often has high maintenance cost that can erase ROI.
Likely follow-up: Why does process volatility hurt RPA ROI?
What the interviewer is checking: Business-value thinking.
Common mistake: Automating low-volume or unstable processes with poor ROI.
Q5. How do you handle changes to underlying applications that break bots?
Short answer: Robust selectors, monitoring, a governance/maintenance process and change awareness.
Strong answer: UI changes are the top cause of bot failure. Mitigate with resilient selectors/anchors, monitoring and alerting on failures, a maintenance/support model, and coordination with app teams about upcoming changes. Treat bots as software needing versioning, testing and ongoing maintenance — not fire-and-forget.
Likely follow-up: Why treat bots as maintained software?
What the interviewer is checking: Operational RPA maturity.
Common mistake: Deploying bots with no maintenance plan.
Q6. Where does intelligent automation / AI fit with RPA?
Short answer: AI (OCR, ML, GenAI) extends RPA to unstructured data and decisions.
Strong answer: RPA handles structured, rule-based steps; adding AI — document understanding/OCR, ML classification, and increasingly GenAI/LLMs — lets automations process unstructured inputs (invoices, emails) and make judgment-based decisions. This "intelligent automation" expands what can be automated beyond deterministic rules.
Likely follow-up: What kind of task needs AI on top of plain RPA?
What the interviewer is checking: Awareness of the 2026 automation direction.
Common mistake: Assuming RPA alone handles unstructured documents well.