Glossary of AI agent, assessment and verification terms.

AI agents

AI agent

Software that uses a language model to decide the next step in a task, calls tools such as a CRM or database, and checks the result. A chatbot replies; an agent also completes work in connected systems. Learn more about AI agent

Agentic AI

An approach where AI systems plan and carry out multi-step work with some autonomy, inside limits set by people. The label is widely overused, so ask what a system actually does without approval. Learn more about Agentic AI

AI employee

Splashgain’s name for a set of agents that covers one job, such as helpdesk or HR operations. It prepares and routes the work; people keep the decisions. Learn more about AI employee

Chatbot

A conversational interface that answers questions, usually from a script or a knowledge base. It does not act in other systems on its own.

Robotic process automation (RPA)

Software that repeats fixed steps across applications, such as copying fields between screens. It is reliable for stable, rule-based tasks and brittle when the input varies.

Human in the loop

A design in which a person reviews or approves an AI system’s output before it takes effect. It is the norm for decisions involving money, contracts or people.

Draft-and-approve

The default rule on Splashgain agents: the agent prepares the work and a named person approves it before anything is sent, paid or signed. Learn more about Draft-and-approve

Retrieval-augmented generation (RAG)

A technique in which a model first retrieves relevant passages from your documents and then answers from them. It keeps answers grounded in your content and lets each answer cite its source.

Model Context Protocol (MCP)

An open protocol, introduced by Anthropic in 2024, for connecting AI applications to tools and data sources through one standard interface.

Guardrails

Rules and checks around an AI system that block unsafe, off-policy or out-of-scope outputs and actions before they reach a customer or a system.

Hallucination

An answer from a language model that sounds confident but is not supported by facts or sources. Grounding answers in documents and requiring citations reduces it.

Lead enrichment

Adding company, role and context details to a sales lead from public and internal sources, so a rep knows who they are contacting and why. Learn more about Lead enrichment

Verification, data and service

KYC (Know Your Customer)

The checks regulated businesses run to verify a customer’s identity and address before providing a service, usually with identity documents and database lookups. Learn more about KYC (Know Your Customer)

Document verification

Checking that a document is genuine and that its details match the person and other records, such as a certificate against the body that issued it. Learn more about Document verification

DigiLocker

The Government of India’s digital document wallet, where issuers such as education boards and transport departments publish documents that can be verified at source. Learn more about DigiLocker

DPDP Act

India’s Digital Personal Data Protection Act, 2023. It governs how organisations collect, use and keep personal data, including consent and purpose limits.

Data fiduciary

Under the DPDP Act, the organisation that decides why and how personal data is processed, and that is responsible for protecting it.

Grievance redressal

A structured process for receiving, tracking and resolving complaints within set timelines, used across government offices and regulated sectors. Learn more about Grievance redressal

Service level agreement (SLA)

The committed response or resolution time for a request, and the measure a helpdesk or office is judged against. Learn more about Service level agreement (SLA)