Select appropriate Claude product features (Projects, research mode, chat, artifacts)
Pick the Claude.ai feature that matches the job: Projects for recurring context, Research for sourced multi-step web work, artifacts for documents, tables and small tools, plain chat for one-off questions.
Key points
- 1
Projects are self-contained workspaces with their own chat history, uploaded knowledge and custom instructions. Knowledge and instructions persist across every chat inside the Project, so recurring context (style guides, handbooks, client briefs, comparison rules) belongs there rather than in every prompt.
- 2
On paid plans, when Project knowledge approaches the context limit Claude switches to retrieval (RAG) and loads only the relevant parts of the knowledge for each question, so a Project can hold far more than a single conversation.
- 3
Research runs multiple searches that build on each other, across the web and connected sources such as Gmail, Google Calendar and Google Docs, and returns a report with easy-to-check citations. Use it for comprehensive, sourced, multi-step information gathering (competitor comparisons, updating internal documents with current external facts). It requires web search to be on and is on paid plans.
- 4
Plain web search is for a quick current fact; extended thinking is for harder reasoning without external sources; Research is for gathering and synthesising across many sources. They can be combined.
- 5
Artifacts are what Claude makes that you would put in front of someone: documents, tables, code, diagrams, single-page web tools, dashboards. Claude creates them for significant, self-contained content (typically over 15 lines) that you will edit, iterate on or reuse outside the conversation. They can be edited in place, versioned and published.
- 6
Artifacts are outputs, not storage: they do not carry context into future chats. Persisting context is a Project's job.
- 7
Plain chat is right for one-off questions and short answers; an artifact for a one-line fact or a Project for a single question is overhead without benefit.
- 8
Features combine: run Research inside a Project so current external facts are compared against persistent internal knowledge; build the deliverable as an artifact in the same Project so it is versioned and shared context.
- 9
Memory records preferences and working context, not verbatim documents; it is not a substitute for Project knowledge when a specific file must be consulted.
- 10
Common distractors: pasting the same document into every chat; keeping one chat open indefinitely; using Research to "look up" your own internal document; expecting an artifact to persist context; relying on memory to hold a 90-page handbook.
Read the source
Test yourself on Select appropriate Claude product features (Projects, research mode, chat, artifacts)
Ten questions, with the answer and explanation after each one.