What is the Claude application?
The Harvard Chan AI Glossary
In our classes, we often pause to explain the ideas behind AI and the language used to describe it. These posts grew out of those conversations. They’re for anyone who would like to understand AI better, with no technical background assumed.
The Claude application is Anthropic’s assistant product for conversation and working with information. It provides a user-facing environment around Claude model capabilities. A model is the trained component that interprets material and generates responses; the application supplies the conversation, files and other available working features.
“Claude” can therefore mean either the model family or the assistant. Claude Code is another related name, referring to the coding-agent product. Distinguishing them helps explain why a task may behave differently in different environments even when the models are related.
You might supply discussion notes to Claude and ask it to develop a proposal. A proposal generated from notes still needs review: it may include assumptions that were never agreed or omit an important qualification.
The term artifact is sometimes used for a document, piece of code or other output produced for inspection and reuse. An artifact is not automatically a finished deliverable. A written plan can be coherent while depending on unverified assumptions. A small application can look complete while missing important behavior.
An application may also connect to tools, meaning software functions for retrieving information or performing actions. The available functions and permissions depend on the setup. A model describing an action does not prove that a tool ran, and a tool reporting success does not prove that the action was appropriate.
Context is the information supplied for the current task. It may include selected notes, earlier messages and instructions. Files stored in a project are not necessarily all present in every response. For an important conclusion, inspect which material actually supports it rather than assuming that a familiar workspace guarantees complete recall.
The class used supplied discussion notes to develop a proposal. A healthcare adaptation would be to provide non-confidential notes from a referral-process workshop and ask Claude to separate agreed requirements, unresolved questions and proposed next steps. Reviewers should check that a participant’s suggestion has not become an approved policy in the resulting draft. The application can help turn discussion into a work product; responsibility for what was actually agreed remains with the people involved.
Source: Anthropic — Claude Product Overview.
If you would like to find out more about AI and improve your leadership skills, consider the AI in Health Care Certificate of Specialization from the Advanced Learning Academy at Harvard T.H. Chan School of Public Health. You can take an individual course or work toward the full certificate.
AI is a fast-moving field. This explanation reflects our class discussions and related material at the time of writing. Terminology, capabilities and tools may change.
For suggested additions or corrections, please email me at tpanch@hsph.harvard.edu.
Dr. Trishan Panch
Program Director, AI in Health Care
Harvard T.H. Chan School of Public Health — Advanced Learning Academy