AI Glossary
Plain language definitions of the enterprise AI terms you meet across the Blits.ai platform, from digital humans and agentic AI to RAG, guardrails and agentic payments.
- Conversational AI
- Technology that lets software hold a natural language conversation with people through text or voice, understanding intent and responding in context rather than following a fixed menu.
- Digital Human
- A lifelike AI avatar that combines a realistic face and voice with conversational AI, so customers can talk to it face to face across web, kiosk or video.
- Voice Bot
- An AI agent that speaks and listens over a phone line or microphone, handling inbound and outbound calls with streaming speech recognition and text to speech.
- Chatbot
- A text based conversational agent on a website, app or messaging channel that answers questions and completes tasks automatically.
- Agentic AI
- AI that does not just answer but acts. An agentic system plans a goal, uses tools and data, takes steps, checks the result and reports the outcome, all under policy and guardrails.
- AI Agent
- A configured unit of agentic AI with a role, a set of tools, memory and permissions, responsible for owning a task from intake to resolution.
- Large Language Model (LLM)
- A model trained on large amounts of text that predicts and generates language, powering understanding, summarisation, reasoning and conversation. Examples include Google Gemini, OpenAI GPT and Anthropic Claude.
- Custom LLM
- A language model adapted to a specific organisation or industry, through fine tuning or grounding, so its answers reflect your data, tone and rules.
- Fine-tuning
- Further training of an existing model on your own examples so it performs better on your specific tasks, terminology and style.
- Retrieval Augmented Generation (RAG)
- A technique that retrieves relevant passages from your knowledge base at question time and gives them to the model as context, so answers are grounded in your documents instead of the model's memory.
- Vector Database
- A store that holds text as numerical embeddings so a system can find passages by meaning rather than exact keywords. It is the retrieval engine behind RAG.
- Grounding
- Tying a model's answers to trusted source material, so responses can be traced back to real documents and are less likely to be invented.
- Hallucination
- When a model produces a confident answer that is factually wrong or unsupported. Grounding, retrieval and guardrails reduce it.
- Guardrails
- Rules and checks that keep an AI system inside safe and on brand behaviour, for example blocking disallowed topics, enforcing tone or requiring human approval for sensitive actions.
- Prompt Engineering
- The practice of designing the instructions and context given to a model to get reliable, high quality output.
- Orchestration
- Coordinating models, tools, channels and steps into one working solution, so the right component runs at the right moment. The Blits.ai platform is an orchestration layer.
- Text to Speech (TTS)
- Technology that converts written text into natural sounding spoken audio, giving voice bots and digital humans their voice.
- Speech to Text (STT)
- Technology that transcribes spoken audio into text so an AI agent can understand what a caller said. Also called speech recognition.
- Model Context Protocol (MCP)
- An open standard that lets AI applications connect to external tools and data sources in a consistent way, so agents can safely reach the systems they need.
- Agentic Payments
- Emerging standards and controls that let an AI agent make or authorise a payment on a user's behalf within strict, auditable limits.