Build a free open-source WhatsApp CRM with Meta's official API. Follow Supabase setup, webhook config, and your first keyword automation.
Most people open Claude.ai, type one question, and call that setup. Then they hit a usage wall by Wednesday, notice Claude has no memory of last week's conversation, and wonder why their carefully written instructions never seem to apply.
None of that is bad luck. It is an incomplete setup.
Claude has a real configuration layer sitting under the chat box: plans, models, effort levels, instructions, projects, skills, memory, and connectors. Most of it takes one click to turn on. Almost none of it is active by default beyond the basics.
This guide walks through that setup in the order that actually matters, starting with the question most people get wrong first.
The right way to set up Claude in 2026 is to pick a plan based on real usage instead of fear of limits, install the web, desktop, and mobile apps so context follows you everywhere, set account-wide instructions once, then turn on memory and start a project. Add skills and connectors only once you have a real task for them. Most setup mistakes come from skipping the personalization layer, not from picking the "wrong" plan.
Anthropic currently offers five consumer and team tiers: Free, Pro, Max 5x, Max 20x, and Team. Enterprise sits above Team for larger organizations with custom seat counts.
Here is the detail almost every comparison page skips. The Max plans do not unlock new abilities.
A Pro subscriber and a Max subscriber can use the same models, the same projects, the same Claude Code access, and the same Cowork features. The only real difference is how much usage you get before you hit a wall.
Upgrading to Max does not make Claude smarter. It makes the conversation longer before you have to stop and wait.
That distinction matters because it is the single most common overspend in this whole setup. People jump to a $100 or $200 plan in week one, before they know whether Pro's limits were ever actually a problem for them.
Here is what each tier includes, based on Anthropic's current plan documentation; plan details may change over time.
| Plan | Price | Usage | Watch Out For |
|---|---|---|---|
| Free | $0 | Session-based limit, resets roughly every 5 hours | No Claude Code or Cowork access. Limits drop further during high-demand periods. |
| Pro | $20/mo | At least 5x Free's session usage, plus a weekly cap | Does not include API usage through the Console. That is billed separately. |
| Max 5x | $100/mo | 5x Pro's usage per session | No new features over Pro. Purely a usage ceiling increase. |
| Max 20x | $200/mo | 20x Pro's usage per session | Priority access to new models and features, but same capability set as Pro. |
| Team | From $20–25/seat | 1.25x–6.25x Pro per seat, depending on seat type | Requires a minimum of 5 seats. Adds admin tools, SSO, and workplace connectors. |
A free account is enough to follow every setup step in this guide. If you use Claude daily for real work, Pro is usually the better starting point, not Max.
If you do decide to pay, an annual subscription bills once a year at a discounted rate compared to paying monthly. Pro, Max, and Team subscribers can also turn on usage credits, which let you keep working past your included limit instead of waiting for a reset, billed only for the extra usage beyond the plan.
Save a Max upgrade for after you have actually hit Pro's weekly or session limit more than once. You can check your own usage and next reset time under Settings > Usage at any time, so you do not have to guess. If usage limits keep blocking you, the dedicated Claude usage limits guide covers how the resets work in more detail.
If you are setting this up for a team Team plans add Single Sign-On, centralized billing, and connectors to Google Drive, Gmail, Calendar, GitHub, Microsoft 365, and Slack that apply across the whole workspace. None of that is available on individual Pro or Max accounts, so do not buy five separate Pro seats expecting team features to appear.Claude runs on the same account everywhere, but each surface unlocks a different habit. Skipping one usually means losing context when you switch devices.
1 ~2 minutesGo to claude.ai and create an account with an email address. You must be at least 18 years old and in one of Anthropic's supported locations. You will know it worked when you land on the chat screen with a model name showing below the message box.
Download the Mac or Windows app from claude.ai/download. The desktop app is what unlocks Cowork and the Claude in Chrome connector later in this guide, so it is worth installing even if you mostly chat in the browser.
3 ~1 minuteDownload Claude for iOS or Android from your app store. On Pro and Max plans, this lets you message Claude from your phone while a Cowork task keeps running on your desktop in the background.
On mobile, the model selector moves to the top of the screen instead of sitting below the input box like it does on web and desktop. Same setting, different spot. That is the kind of small layout difference that trips up beginners switching between devices for the first time.
One more thing worth checking before you build any real workflow around Claude: confirm you are in a supported location. Anthropic maintains a current list, and access can vary slightly by country even within otherwise supported regions.
Every message you send picks three things at once: a model, an effort level, and whether extended thinking is on. Most people never touch any of them.
Click the model name next to the send button to open the picker. Opus is the most capable and slowest option, Sonnet balances speed and quality for most everyday tasks, and Haiku is built for speed on simple jobs. Model names get periodic version updates, so check the picker itself for the exact current lineup rather than trusting a name printed in an older guide.
Each model also shows a recommended effort level marked "Default" in the menu.
| Effort | Best for | Usage cost |
|---|---|---|
| Low | Quick questions, simple lookups | Stretches usage furthest |
| Medium | Routine writing and everyday tasks | Light |
| High | Best overall balance of quality and speed | Moderate |
| Max | Genuinely difficult, longer-running problems | Heaviest |
Extended thinking is a separate toggle from effort, switched on or off next to the model name. You can pair any model with any effort level and either thinking setting, so do not assume a model's name alone decides how thorough an answer will be.
This is the step almost everyone skips, and it is the one that actually makes Claude feel like it knows your work instead of starting from zero every time.
Claude has three personalization layers, and beginners regularly mix them up: account-wide instructions, project instructions, and skills.
1 ~3 minutesClick your initials in the lower left corner, choose Settings, then find "Instructions for Claude." Anything you write here applies to every conversation you start, inside or outside a project.
I'm a [your role]. I prefer [tone, e.g. direct and concise]. Default to [format, e.g. short paragraphs over long ones]. If you are uncertain about something, say so instead of guessing.Paste something like that, adjust the brackets, and save. You will know it worked the next time you open a brand new chat and Claude already responds in that tone without you repeating yourself.
2 ~3 minutesProjects are available on every plan, including Free, which can create up to five. Click "Projects" in the sidebar, then "Create project," and give it a name tied to one real piece of work, not a vague catch-all.
This project is for [specific goal, e.g. drafting my weekly client newsletter]. Use [tone/structure]. Always check [specific detail, e.g. the brand glossary I uploaded] before suggesting wording.Project instructions only apply inside that one project, separate from your account-wide instructions. On paid plans, project knowledge scales automatically through Retrieval-Augmented Generation once you upload enough documents, expanding usable capacity by roughly 10x without you doing anything extra.
3 ~2 minutesClick "Customize" in the sidebar, open the Skills tab, click the "+" button, then "Browse skills." Install anything relevant. It turns on automatically and you can toggle it off later without losing it from your list.
If an older guide says "Styles" Styles are migrating into Skills, according to Anthropic's official skills documentation. If a guide or video tells you to set up a custom Style and you cannot find that menu, that is why. Go to Customize > Skills instead, and any style-like behavior you want lives there now.Skills come in a few flavors worth knowing apart. Anthropic skills, like enhanced document creation for Word, Excel, PowerPoint, and PDF files, are built in and load automatically.
Custom skills are ones you or your organization write for repeatable workflows. Partner skills, from companies like Notion, Figma, and Atlassian, are built to pair with their own connectors. The Claude Skills automation guide goes deeper into building your own.
If you are on a Team or Enterprise plan, projects can also be shared with colleagues at different permission levels, from view-only access to full editing rights, as described in Anthropic's projects documentation. That collaboration layer sits on top of everything in this section, not in place of it, so set up your own instructions and projects first either way.
Memory and "search past chats" are not the same feature. Mixing them up is one of the most common setup mistakes people make.
Search and reference chats lets Claude pull specific details from old conversations when you ask it to. It only runs on Pro, Max, Team, and Enterprise plans.
Memory is different. It quietly builds a running summary of your role, preferences, and ongoing work in the background, and it is available to every plan, including Free.
| Feature | Free | Pro / Max | Team / Enterprise |
|---|---|---|---|
| Memory (background summary) | ✓ | ✓ | Owner-controlled |
| Search and reference past chats | — | ✓ | ✓ |
| Incognito chats | ✓ | ✓ | Still exported |
| Per-project memory space | ✓ | ✓ | ✓ |
Turn memory on under Settings > Capabilities, in the Preferences section. If you ever want a clean slate without losing memory permanently, choose Pause instead of Reset. Reset deletes every memory, including project memories, and cannot be undone, per Anthropic's memory documentation.
You will know memory is working when Claude references something from a past conversation without you re-explaining it, or when you open "Manage memory" in the same settings panel and see a real summary instead of an empty list.
For anything sensitive Click the ghost icon before you start a new chat outside a project to open an incognito chat. It is available on every plan, and Claude will not search it or remember it later. Close it when you want normal memory back on.Memory is meant for ongoing work context like your role, communication style, and project details. Anthropic explicitly recommends against treating it like a vault. Do not ask Claude to remember passwords, account numbers, or anything you would not want stored in a third-party system.
Connectors let Claude read from, and in some cases act on, services you already use instead of you copying and pasting context by hand every time.
Open Customize in the sidebar, switch to the Connectors tab, and click "Connect" on anything relevant to your work. Team and Enterprise plans add workplace-wide connectors for Google Drive, Gmail, Calendar, GitHub, Microsoft 365, and Slack, plus an enterprise search project that indexes them automatically for everyone in the organization.
For deeper connections, like an internal database or a custom tool, Claude also supports MCP, the Model Context Protocol. That setup involves a configuration file rather than a single click, so it gets its own walkthrough in the Claude MCP setup guide.
Once the chat setup is solid, three more capabilities sit on top of the same Pro, Max, Team, or Enterprise plan. None of them need a separate purchase.
Claude Code is a terminal-based coding agent that also runs inside VS Code, Cursor, and JetBrains IDEs. It shares the same usage limits as your regular chats. The full walkthrough lives in the Claude Code setup guide, but the short version is: install it, sign in with your existing Claude login, and it just works.
Watch for this If you have an ANTHROPIC_API_KEY environment variable set on your machine from an earlier project, Claude Code will quietly use that key for billing instead of your subscription. That means pay-per-token API charges instead of usage already covered by Pro or Max. Remove the variable if you want Claude Code to use your plan's included usage.Cowork runs inside Claude Desktop and takes on multi-step tasks without you replying to every message along the way. It needs the desktop app open and your computer awake to keep working, and its memory only persists inside projects, not in standalone sessions.
Computer Use lets Claude operate a browser or desktop directly on your behalf, including through the Claude in Chrome connector covered earlier. The Computer Use beginner guide walks through safe first tasks to try before handing over anything sensitive, like a payment form or an account with financial access.
There is also a newer addition worth knowing about: Claude Design, currently in beta on Pro, Max, Team, and Enterprise plans. It opens a chat-plus-canvas workspace at claude.ai/design for building interfaces that can sync directly into Claude Code when a design is ready to become real software. It is not required for the setup in this guide, but it is worth bookmarking if your work involves UI at all.
All four use the same account, the same plan, and the same usage limits you already set up earlier in this guide. There is nothing extra to configure beyond what you have already done.
A short list of what actually trips people up, based on the patterns above:
None of these are hard to fix. Most take under five minutes once you know where to look, which is the entire point of going through the setup in order instead of skipping straight to the chat box. Once your account is configured, the Best AI Research Tools guide covers tools that pair well with a properly set up Claude account.
Start on Free or Pro, not Max. Most people overestimate how much usage they need before they have even tried the personalization features that make each session more useful in the first place.
The setup that matters most is not the plan tier. It is the fifteen minutes spent writing account-wide instructions, starting one real project, and turning memory on. That is what makes Claude feel different from a search box, not the price you pay.
Be careful with two specific defaults: leaving effort on Max for everyday replies, and an old ANTHROPIC_API_KEY variable quietly billing your API account instead of your subscription. Both are easy to fix once you know they exist.
The safest next step is the smallest one. Open Settings, write three sentences about how you want Claude to respond, then start your first project from there.
Skip straight to the part that actually changes how Claude feels to use. Write your instructions, start one project, and turn memory on. Everything else in this guide can wait until you have a real reason to need it.
AI App Studio shares AI tools, app reviews, and practical insights — discover what to use, how it works, and what's new in AI.