Glossary / Artificial intelligence
AI agent
Also called: autonomous agent
An AI agent is a program that receives a goal, decides which steps to take and acts inside tools to achieve it.
A chatbot answers a question. An agent goes further: it chains actions together. It reads an email, looks up information in a file, fills in a record, sends a message, then checks the result.
It relies on an LLM to understand and decide. It is given a goal, the tools it can use and some limits. The more sensitive the task, the more a human should approve before the final action.
An example
Every morning, an agent reads the new requests received, sorts them by urgency, prepares a draft reply and creates the task in the management tool.
AR Digital’s advice
Start with an agent that prepares and let a human approve. Widen its autonomy once it has proved itself.
AR Digital / Artificial intelligence
In the same family.
- ChatGPTChatGPT is the generative AI assistant from OpenAI. It answers questions, writes, summarises, translates and analyses documents.
- Generative AIGenerative AI is artificial intelligence capable of producing new content: text, images, sound, video or code.
- HallucinationA hallucination is a false or invented answer that an AI presents with confidence, as if it were true.
- Local AIA local AI is an artificial intelligence model that runs on the company’s own machines, without sending data outside.
- Machine learningMachine learning is the branch of AI where a program learns from examples instead of following rules written by hand.
- RAGRAG is a method that makes an AI look up the right passages in your documents before writing its answer.




