Claude Setup Guide: 14 Settings & Features That Turn Default Claude Into a Power Tool
Most people use Claude straight out of the box. Here's the complete 14-feature guide — Web Search, Skills, Projects, Cowork, Dispatch, Scheduled Tasks, Artifacts, and more — to set it up properly, once.
Most people open Claude, type a question, and never touch a single setting. That works — but it leaves most of what Claude can actually do switched off. This guide covers all 14 features across two areas: the 8 one-time settings that change how Claude searches, remembers, and reasons, plus 6 productivity tools — Cowork, Dispatch, Scheduled Tasks, Artifacts, and more — that turn Claude from a chat window into something that gets real work done on its own.
What "Claude Basic Setup" Actually Means
Claude basic setup is the process of turning on the features that are off by default — live web search, custom instructions, saved processes, connections to apps like Gmail or Drive, and memory across conversations — so every new chat starts with more context instead of a blank slate. On top of that, Claude also ships with a set of productivity tools most people never open — ways to delegate whole tasks, work from your phone, automate recurring work, and turn results into shareable reports. None of the 14 items below cost anything extra on top of your plan; they're built into Claude and just need to be switched on once.
The 8 Core Settings Worth Turning On
Set these up roughly in this order — each one builds on the last.
Turn on Web Search (and Research)
By default, Claude answers from what it already knows, which has a training cutoff. Click the "+" next to the message box and enable Web Search — for anything time-sensitive (prices, current events, recent releases), this is the difference between a guess and a checked answer.
Move from the browser to the desktop app
The desktop app carries more capability than the browser tab — local file access, connector integrations, and a more stable long-running session. If you use Claude daily, the desktop app is worth the one-time install.
Turn a repeated process into a Skill
If you find yourself typing the same instructions every time — "format this as a table," "review this the way I like it" — save it as a Skill once. After that, you invoke it by name instead of re-teaching Claude the same process in every conversation.
Write your preferences into global instructions, once
Custom instructions (set once in your settings) tell Claude how you like answers formatted, what tone to use, and what to always check before responding. Set them once and every future chat inherits them — no more repeating yourself at the top of every prompt.
Match the reasoning effort to the task
Claude lets you adjust how much it "thinks" before answering. Low effort is fine for quick, simple questions. For anything with several moving parts — planning, analysis, debugging — raising the effort level gives Claude room to reason through the problem before committing to an answer.
Pick the right model for the job, not just the default
Claude isn't one model — it's a family. Haiku is built for speed on simple tasks, Sonnet is the balanced everyday choice, Opus is for the hardest reasoning work, and Fable is a newer option worth trying for tasks that reward a different style of response. Match the model to what you're actually asking for.
Connect the apps you already work in
Connectors let Claude read directly from Gmail, Slack, Google Drive, and similar tools — with your permission — instead of you copy-pasting context into the chat every time. Set these up once in your settings and Claude can pull in the relevant document, thread, or email on its own.
Turn on chat memory and build a Project
"Search and reference past chats" lets Claude recall context from earlier conversations instead of you re-explaining it. Projects go a step further — bundle your files and instructions once, and every chat inside that Project automatically carries that context with it.
6 More Ways to Work Smarter With Claude
Once the basics above are switched on, these six push Claude from a chat window into something that plans, automates, and delivers finished work with far less hand-holding.
Use Cowork to hand off entire tasks, not just questions
Delegate complex, multi-step work to Claude through Cowork. It breaks the task down, executes each part, and delivers a finished result — start to finish, autonomously — instead of you managing it prompt by prompt.
Point Claude at a folder and let it work in your files directly
Give Claude a project folder and it can read, understand, and edit your documents and code right there — no more copying content in and out of the chat to get something reviewed or updated.
Use Dispatch to run Claude from your phone
Give Claude instructions, receive updates, and keep work moving — even when you're far from your desk. Dispatch means being away from your laptop no longer means the work stops.
Automate recurring work with Scheduled Tasks
Set a task up once and Claude runs it on your schedule automatically — a weekly report, a recurring review, a regular check-in. You stop remembering to ask; Claude just runs it.
Turn data into reports and dashboards with Artifacts
Instead of a wall of plain text, Artifacts turn Claude's output into beautiful, interactive reports and dashboards you can actually use and share with others.
Talk to Claude instead of typing, with Voice Mode
Speak naturally to Claude and it listens, understands, and responds — hands-free. Useful for thinking out loud, or any time typing isn't convenient.
Default Claude vs. a Properly Set Up Claude
The model is the same either way — what changes is how much context and capability it's actually allowed to use.
| Setting | Left on default | After basic setup |
|---|---|---|
| Current information | Answers from training data only | Checks live sources via Web Search |
| Repeated instructions | Re-typed in every chat | Saved once as a Skill or global instruction |
| Reasoning depth | One fixed effort level | Matched to task difficulty |
| Your workspace context | Pasted in manually each time | Read directly via connectors |
| Conversation history | Forgotten after each chat | Referenced automatically, or carried in a Project |
| Multi-step work | Managed through back-and-forth prompts | Delegated whole to Cowork and delivered finished |
| Your files | Copy-pasted into the chat one at a time | Read and edited directly from the folder |
| Being at your desk | Required to give Claude any instruction | Optional — Dispatch runs it from your phone |
| Recurring tasks | Manually re-run every time | Automated on a schedule |
| Sharing results | Plain text you reformat yourself | Interactive reports and dashboards via Artifacts |
One trade-off worth knowing: higher reasoning effort and larger models take longer to respond and use more of your usage limits. Reserve the highest effort and the biggest models for genuinely hard tasks — for quick lookups and simple edits, the fast, lightweight settings are the better choice.
Key Takeaways
Claude's most useful features — web search, memory, connectors, Skills — are off by default and take a few minutes to switch on.
Set global instructions and Skills up once, so you stop re-explaining your process in every new chat.
Match reasoning effort and model choice to the task — save the heaviest settings for genuinely hard problems.
Connectors and chat memory remove the need to keep pasting the same context into every conversation.
A Project bundles your files and instructions once, so every chat inside it starts with full context automatically.
Cowork and Scheduled Tasks let Claude run multi-step or recurring work on its own, instead of you prompting it through every step.
Point Claude at a folder for direct file access, use Dispatch to work from your phone, and turn output into shareable dashboards with Artifacts.
Frequently Asked Questions
What is Claude basic setup?
Claude basic setup is switching on the features that are off by default — web search, custom instructions, Skills, connectors, and chat memory — so every conversation starts with better context instead of a blank slate. It's a one-time setup, not something you repeat per chat.
Do I need a paid plan to use these settings?
Web search, global instructions, and Skills are available to most users. Connectors, Projects, Cowork, Dispatch, and higher usage limits on the larger models are typically tied to your specific plan — check your account settings to see what's available on yours.
What's the difference between Claude's models — Sonnet, Opus, Haiku, and Fable?
They're built for different jobs: Haiku is optimized for speed on simple tasks, Sonnet is the balanced default for everyday work, Opus is for the most demanding reasoning tasks, and Fable is a newer model worth testing for tasks where its response style fits better. Picking the right one for the task, rather than always using the default, is part of a proper setup.
What are Claude Skills, and how are they different from custom instructions?
Custom instructions are always-on preferences that apply to every chat. A Skill is a specific, repeatable process — like a formatting template or a review checklist — that you invoke by name only when you need it, instead of it running in the background of every conversation.
What are Claude connectors?
Connectors let Claude read from apps you already use — like Gmail, Slack, or Google Drive — with your permission, so it can pull in a relevant email, document, or message directly instead of you copying and pasting it into the chat.
What is Claude Cowork?
Cowork is how you delegate a real, multi-step task to Claude instead of walking it through each step yourself. You describe the outcome you want and Claude breaks it down, executes it, and delivers the finished work.
What is Dispatch used for?
Dispatch lets you instruct Claude and check on progress from your phone. It's built for staying in the loop and keeping work moving when you're away from your desk, not just for quick chats.
How do Scheduled Tasks work in Claude?
You set up a task once — what to do and how often — and Claude runs it automatically on that schedule going forward, without you needing to remember to ask again.
What are Claude Artifacts?
Artifacts are a way for Claude to turn its output into an interactive report, dashboard, or document instead of a plain text reply — something you can view, use, and share on its own.
More Claude Tips Coming Soon
This post covers Parts 1 & 2 of the Claude Learning series — the core settings and the productivity tools. Follow along for what's next.
Follow on Instagram →