Automation & AI glossary
Plain-language definitions of the terms that come up in automation and AI projects. Each entry links to the service that puts it into practice.
- AI Automation
- The use of machine learning and language models to carry out business processes that previously required human judgment, such as classifying documents, extracting data, and routing decisions.
- Robotic Process Automation (RPA)
- Software robots that perform rules-based tasks across existing applications by interacting with them the way a person would — logging in, reading screens, and moving data — without changing the underlying systems.
- AI Agent
- Software that works toward a goal rather than following a fixed script: it plans steps, uses tools and systems, and applies judgment within the permissions you define.
- OCR (Optical Character Recognition)
- Technology that converts scanned documents, PDFs, and images into machine-readable text. OCR reads characters but does not, on its own, understand what the fields mean.
- Arabic OCR
- OCR tuned for Arabic script, which is written right-to-left with connected letterforms, context-dependent character shapes, and optional diacritics — features that cause general-purpose engines to misread text.
- Intelligent Document Processing (IDP)
- The combination of OCR, AI, and business rules that turns unstructured documents into validated, structured data ready for a downstream system.
- Document Understanding
- AI that identifies a document's type, extracts the fields that matter, validates them against existing records, and determines the next action — going beyond reading text to interpreting it.
- Straight-Through Processing (STP)
- The share of transactions completed end to end automatically, with no human touch. A common measure of how effective an automation actually is.
- Human-in-the-Loop
- A design in which automation handles the cases it is confident about and routes uncertain ones to a person, whose corrections can be fed back to improve accuracy over time.
- Exception Handling
- The rules and routing that decide what happens when a transaction cannot be processed automatically — for example a mismatched invoice or a low-confidence extraction.
- Workflow Orchestration
- Coordinating tasks, approvals, systems, and people so a process runs in a defined order with tracked state, instead of through ad-hoc email and spreadsheet handoffs.
- Business Process Automation (BPA)
- Automating an entire business process end to end — including approvals, integrations, and exceptions — as opposed to automating a single isolated task.
- Three-Way Match
- An accounts payable control that compares the purchase order, the goods receipt, and the supplier invoice before payment is approved, preventing overpayment and duplicate billing.
- Procure-to-Pay (P2P)
- The complete purchasing cycle, from purchase request and purchase order through goods receipt, invoice, matching, and supplier payment.
- Hire-to-Retire (H2R)
- The complete employee lifecycle, from hiring and onboarding through transfers and promotions to offboarding and final settlement.
- Employee Self-Service (ESS)
- A portal or workflow where employees request letters, update their details, and submit leave themselves, with each request routed for approval and recorded automatically instead of being emailed to HR.
- Legacy Modernization
- Extending or progressively replacing ageing systems that still run critical work — usually by automating and integrating around them rather than attempting a single high-risk replacement.
- ERP Integration
- Connecting an ERP system with other business applications so data flows between them automatically while the ERP remains the authoritative record.
- System of Record
- The application designated as the authoritative source for a given data domain. Other systems may hold copies, but conflicts are resolved in favour of the system of record.
- API Integration
- Connecting systems through their published programming interfaces, which is more reliable and maintainable than automating through the user interface when an API is available.
- Digital Transformation
- Changing how an organization operates by applying digital technology to its processes, data, and services — measured by operational outcomes rather than by the technology deployed.
- Process Mining
- Reconstructing how a process actually runs by analysing event logs from the systems that support it, which reveals variations and bottlenecks that written documentation usually misses.
- Audit Trail
- A complete, timestamped record of who or what performed each action in a process, required for compliance reviews and for diagnosing exceptions.
- Large Language Model (LLM)
- An AI model trained on large volumes of text that can interpret and generate language. In automation it is typically used for classification, extraction, summarization, and drafting, inside defined guardrails.