You have installed Claude Desktop. The app is open. You signed in. There is a chat window in front of you and you are not sure what to do first. This guide gives you a structured 30 minutes that takes you from "blank app" to "actively useful" — without skipping anything important and without padding the time.
Work through the steps in order. Each one builds on the last.
Minute 0–5: A first conversation that proves it works
Before configuring anything, send one message that uses Claude for something real. The point is to get past the first-message anxiety and see how Claude actually responds.
Pick something small from your actual work. Examples: "Summarise the key obligations of an Irish employer under the Organisation of Working Time Act, 1997 — for someone new to HR." Or: "I have a meeting tomorrow with a difficult client. Help me draft three opening lines that are firm but not aggressive."
Send it. Read the response. If it is useful, ask a follow-up. If it is not, refine the prompt and try again. The goal of this five minutes is just to feel what asking Claude is like, with no setup overhead.
Minute 5–10: Create your first Project
The most useful Desktop feature for ongoing work is Projects. A Project is a named container that holds shared context — files and instructions — that every conversation inside it inherits automatically.
- In the left sidebar, click New Project
- Name it something concrete: "Client Work", "Research — Leadership", "Personal Assistant"
- In the project description / instructions field, write 3–5 sentences explaining what this project is for and how Claude should behave in it. For example: "This project is for managing communication and deliverables for my freelance HR consulting clients. Default to UK English. Match a professional but warm tone. Always ask clarifying questions before drafting client-facing content."
- Click save
You now have one Project. Its instructions apply to every conversation you start inside it.
Minute 10–15: Attach a file and use it
Inside the Project you just created, click into the project's files area and upload one document — something you actually work with. A PDF report, a Word doc with your bio, a spreadsheet, a brand voice guide. Anything that gives Claude useful context for tasks in this project.
Once attached, start a new conversation inside the Project. Ask Claude something that requires the attached file: "Based on my brand voice document, draft a LinkedIn post about [topic]." Or: "Pull the three most important obligations from the contract I just attached."
Two things should be clear from the response: Claude actually read the file (it should reference specific content from it), and Claude is following the project instructions you set in the previous step. If the response feels generic or ignores the file, the instructions or the prompt need sharpening.
Minute 15–20: Turn on Memory and watch it learn
Open Settings (gear icon, top right) and find the Memory section. Turn it on if it is not already.
Memory is different from Project instructions. Project instructions apply only inside that project. Memory applies everywhere in Claude Desktop — it is a layer of facts about you personally that Claude carries between every conversation.
To trigger a memory write, tell Claude something explicitly: "Remember that I prefer responses under 200 words unless I ask for more detail." Or: "Remember that I work in Ireland — default to UK English and Euro pricing when relevant."
Claude will confirm it has saved this. To verify: open Settings → Memory and you should see your entry listed. You can edit or delete entries from there at any time.
Minute 20–25: Try the file-upload workflow on something real
Without a Project — just in a regular conversation — drag a file from your computer directly into the message box. Try a PDF that you actually need to deal with: a contract, a long report, a research paper, a complex email thread saved as PDF.
Ask one of:
- "Summarise this in five bullet points."
- "What are the three things I should pay closest attention to in this document?"
- "Extract the deadlines and obligations into a table."
- "What is missing from this document that I would expect to see for [purpose]?"
This is the most common Desktop workflow for knowledge workers — getting fast leverage from documents you would otherwise have to read in full. It works well for documents up to a few hundred pages; very long documents may need to be split.
Minute 25–30: Set yourself one experiment for the week
The biggest mistake at this point is to use Claude for one impressive thing today and then forget about it. The way Claude becomes part of your workflow is to commit to one specific use case for a week and see what happens.
Pick one of these (or invent your own) and put it in your calendar:
- Morning email triage. Each morning for a week, paste your top three unread emails into Claude and ask for a one-line summary and recommended response of each. After a week, decide if it saved time.
- Document-of-the-day. One PDF you would normally avoid — drag it in and ask for a summary before reading the full thing.
- Pre-meeting prep. Before each meeting this week, give Claude the agenda and your past notes from this person; ask what to anticipate and what questions to prepare.
- End-of-day brain dump. Talk to Claude for ten minutes at end of day about what happened and what you are stuck on. Use Memory so context carries between days.
Whichever you pick, the key is repetition for at least a week. One use proves Claude can do it. Five uses tells you whether it actually changes how you work.
What comes next
- Projects and Memory in Claude Desktop — go deeper on the two features you just touched
- Connecting MCP Servers to Claude Desktop — when you want Claude to read your calendar, email, or other live tools
- Cowork: Scheduling Recurring Tasks — for tasks you want Claude to run automatically
- How to Talk to Claude — Prompting Basics — the four-part recipe for better answers