AI Agent
Software that can interpret context and call approved tools to advance a business workflow under guardrails, with escalation to humans for exceptions and high-risk actions.
Canonical definitions used across Tapti Services content. Each term is schema-enabled and linked to related concepts, services, and insights for humans and AI systems.
This glossary defines Enterprise AI, AI Integration, AI Agent, AI Copilot, Document Intelligence, OCR, RAG, MCP, Enterprise Search, Workflow Automation, ERP, CRM, HRMS, and related terms consistently for citation.
Software that can interpret context and call approved tools to advance a business workflow under guardrails, with escalation to humans for exceptions and high-risk actions.
A permission-aware assistant that helps people find grounded answers and draft work inside enterprise applications, with citations and accountable ownership. It assists humans; it does not replace the system of record.
Classification, extraction, validation, and routing of documents into business systems — beyond OCR — so documents become usable operating data with human review for exceptions.
AI capabilities applied inside business systems of record with identity, audit, evaluation, and human oversight for operational and regulated environments.
Connecting models, retrieval, copilots, and agents to existing enterprise applications through APIs and governed workflows, without requiring a full system replacement.
Retrieval-Augmented Generation
A pattern that retrieves relevant passages from approved knowledge sources and provides them as evidence for a model response. In enterprises, RAG should enforce access control, cite sources, and escalate when evidence is weak.
Model Context Protocol
A protocol pattern for wiring assistants and agents to external tools and data sources in a structured way, so tool use can be governed, inspected, and constrained by application policy.
Optical Character Recognition
Converts document images into machine-readable text. OCR alone does not classify documents, validate fields against business rules, or commit structured data to systems of record.
Repeatable, policy-bound process execution across systems — appropriate when steps, rules, and outcomes are well defined and must remain auditable.
Structured workflows, approvals, and integrations that reduce manual coordination while preserving policy control and auditability.
Permission-aware discovery across company documents and records. Modern implementations often combine keyword and semantic retrieval and underpin copilots and RAG.
A structured network of entities and relationships used to organize meaning across content and systems. On this website, the knowledge graph connects services, topics, industries, and definitions for humans and AI systems.
Retrieval that matches meaning as well as exact keywords, typically using embeddings alongside traditional ranking. Best used with permission filters and hybrid keyword search for identifiers.
Connecting applications through well-defined interfaces so data and actions can be exchanged safely. Enterprise AI integration depends on APIs (or events) into systems of record with authorization and validation.
Sequenced improvement of processes, data, and systems so technology supports measurable business outcomes — often through integration and modernization rather than wholesale replacement.
Enterprise Resource Planning
Systems that hold core operational transactions for finance, inventory, manufacturing, and related processes. ERP remains a primary system of record when AI is integrated.
Customer Relationship Management
Systems that manage leads, accounts, opportunities, and customer interactions as systems of record for revenue and service processes.
Human Resource Management System
Systems for employee records, attendance, payroll coordination, and people processes. Often integrated with identity, finance, and workflow systems.
Reporting and analytics that turn operational data into shared views of performance, risk, and demand for leaders and teams.
An approach to building and operating applications that uses cloud delivery models, automation, and scalable services. Enterprise programs may be cloud-native, hybrid, or on-premise depending on residency and risk.
Long-lived business applications that support core processes — including ERP-style systems, CRM, HRMS, portals, and custom platforms — with maintainable architecture and operational readiness.
Adapting a model by further training on domain-specific data. Useful for style or specialized recognition tasks; often complementary to RAG for factual grounding from changing enterprise content.
A conversational interface that answers or routes user messages. Basic chatbots follow scripts or FAQ matching; advanced assistants may use retrieval and tools but are not automatically full AI agents.
Practices and tooling that connect development and operations — CI/CD, environments, observability, and release discipline for production systems.