Beyond the core path
Follow your curiosity.
Explore related ideas when you need them. Each word links to its definition in the living glossary.
Concept explainers
Untangle similar ideas.
Some words sound interchangeable. These comparisons explain how they fit together and when the difference matters.
Memory vs context window
The context window is what the model can work with right now. Memory is saved information the app can retrieve later. Saved memories still need to be brought into context to influence a response.
Think of notes in a notebook versus the pages open on your desk.
Model vs agent vs harness
The model generates outputs. An agent uses a model to choose and carry out steps toward a goal. The harness supplies the tools, context, permissions, and loop that make this possible.
For a report: the model drafts text; the agent decides what to research; the harness runs its searches and returns the results.
Prompting vs RAG vs fine-tuning
Prompting gives instructions. RAG retrieves relevant material to add to the request. Fine-tuning changes the model through additional training.
Ask for your company’s writing style, retrieve its guidelines, or train on examples of that style: three different approaches.