- AI agent
- A program built on a language model that understands a request, decides what to do and uses tools (calendar, CRM, email) to do it, with a person overseeing what matters.
- Flow
- A step-by-step diagram of how an agent acts: what it recognizes, what it decides, which tools it uses and when it hands over to a person.
- Guardrails
- Rules and limits that keep the agent within what it should do: allowed topics, data it must not give out, when it must ask for help.
- Hallucination
- An answer that sounds convincing but is false or not backed by the available information. It is reduced with good sources, clear limits and testing.
- Human in the loop
- A design in which a person reviews or approves what matters and receives the case when the agent should not resolve it alone.
- Knowledge base
- The set of documents and data (catalogs, rates, procedures) the agent relies on to answer.
- LLM (large language model)
- An AI model trained on large amounts of text, able to understand and generate natural language. It is the engine of most current agents.
- MCP
- An open protocol that lets an AI model connect to tools and data sources in a standard way.
- Orchestration
- Coordinating several steps, agents or tools to complete a task from start to finish.
- Prompt
- The instructions and context given to a language model to guide its behavior.
- RAG
- A technique in which the agent first searches your documents and answers based on what it found, instead of relying only on what the model knows.
- Use case
- A specific situation where AI adds value, with a measurable outcome. Each use case is tackled as an independent project.
- Voice agent
- An AI agent that talks on the phone in real time, on inbound or outbound calls.