What is a Chat?

In Goodweek, a Chat is the space where you interact with a language model (LLM). Its primary purpose is to turn an intention, a question, a wording request, or an analysis into an answer generated by the model, using what Goodweek makes available.

This addresses a simple problem: when information is scattered (documents, a Space’s knowledge, data from connected tools), it’s costly to gather everything and explain it from scratch every time. Chat acts as a convergence point: it combines your request with the available contextual elements to generate a more relevant response than if the model had to “guess”.

Each message you send in a Chat is a prompt: it’s not just a sentence, but an instruction that guides the model’s reasoning (goal, constraints, expected format, level of detail). That’s why two different prompts about the same topic can produce very different responses.

A Chat is always associated with an active model, the one you are interacting with at a given moment. Depending on the need, Goodweek may call on another model within the same Chat, for example when certain capabilities (writing quality, speed, specialization) are better suited to a particular request.

Finally, a Chat can draw on several “sources” of understanding:

• The context provided by your messages and attached files

• A Space’s knowledge base

• Information accessible via integrations

• The instructions and rules specific to the Space where the Chat takes place