← Back to all free guides

Welcome to the
Era of
Vibe Working

Claude isn’t only a chatbot anymore. It’s an agent. The people who learn to delegate to it, instead of only talking to it, are about to look untouchable at their jobs.

You’ve heard of vibe coding: you describe what you want and AI builds the software. In February 2026, Anthropic’s head of product for enterprise, Scott White, told CNBC, "I think that we are now transitioning almost into vibe working." (CNBC) Same idea, for every job.

Stop managing tasks and start delegating outcomes. Instead of using Claude as a search engine or writing assistant, you hand it a real piece of work with full context and a clear finish line. It plans the steps, uses your tools and files, and you come back to something finished.

The Shift

01

How most people still use it

"Write me an email to my client about the project delay." Claude writes it, you copy it, you send it. One question, one answer, and you’re still doing all the coordination.

02

How it works now

"Here’s the full situation, here’s what needs to happen, here’s everything you need. Go." Claude reads your project files, drafts the email, checks your calendar and preps the follow-up while you move on.

03

The same request, done both ways

Before: "Write a follow-up email to the client after our call."

After: "Using the call summary and our proposal, draft a follow-up that confirms what we decided and sets next steps with owners and dates."

Outcome-Delegation Prompt
Here's the call recording summary, the client's original brief, the proposal we sent, and the three open issues from today's call. Draft a follow-up email that: - confirms what was decided - addresses each open issue with the resolution we discussed - sets clear next steps, each with an owner and a date Tone should match how I write: direct but warm. Lead with the key point. If anything in the notes contradicts the proposal, flag it to me above the draft instead of guessing. Save the draft to /drafts and don't send anything.

The Rule Of Thumb

Before you send a prompt, ask yourself what finished looks like. If your prompt can’t answer that, Claude can’t either. Give it the full picture once and walk away.

The Three Things To Set Up

01

Projects: Claude knows you before you say a word

Most prompts are long because you re-explain who you are every time. A Project holds your context across every chat inside it.

  1. In claude.ai, open Projects in the sidebar and click + New Project. Name it how you think about your work (Client Work, Content, Operations).
  2. Click Set project instructions and write 5 to 10 sentences covering your role, your style, what the Project is for and your standing rules.
  3. Click + in the project knowledge panel and upload 3 to 5 pieces of work you’re proud of, plus templates or brand guidelines. Claude learns your style from examples faster than from descriptions.

Projects work on every plan, including Free (up to five projects there). (Projects help)

Example Project Instructions
I'm a Senior Marketing Manager at a B2B SaaS company. I manage a team of 4. My communication style is direct, with no filler and no corporate speak. This Project is for client-facing work. Always use formal language in client communications. When writing emails, lead with the key point. Never include more than 3 action items in a single message. When you're unsure about tone, ask before drafting.

02

Connectors: Claude works inside your actual tools

Without connectors you paste in emails and describe project status. With them, Claude reads from and acts in Gmail, Google Drive, Notion, Slack and more.

  1. Go to Customize > Connectors (Customize is being renamed Plugins, so look there if you see that name). You can also browse claude.ai/directory.
  2. Connect your two most-used tools first: email and wherever your docs live. Sign in and approve access.
  3. In any chat, click + and hover Connectors to switch them on or off.

Gmail, Calendar and Drive are available on every plan. (Connectors help)

Try This Once You’re Connected
Check my Gmail and read the Notion board for Project Phoenix. Give me a 5-bullet briefing: 1. Where things stand 2. What needs a response from me today (sender, subject, why) 3. Whether we're on track for the Friday deadline, and what could slip 4. Anything someone is waiting on me for 5. The one thing I should do first Only use what you actually find in Gmail and Notion. If something is unclear, say "unclear" instead of guessing.

03

Delegated tasks: Claude works while you’re gone

Chat answers and waits. Cowork makes a plan and carries it out, reading and writing real files and running on a schedule. On Pro, Max and Ultra, Cowork is being folded into regular chat, so you may simply describe the task in any conversation. If you still see a Cowork option at the bottom left of the message box, pick it.

  1. For work on files on your computer, use the Claude desktop app and connect a single project folder, not your whole drive.
  2. Describe the outcome. Name the file that should exist when it’s done and what goes in it.
  3. Review the plan it proposes, approve it, then walk away.
  4. For recurring work, open Scheduled in the sidebar, or just say how often it should run in the chat.

Cowork is on all paid plans in the desktop app, and on web and mobile for Pro, Max, Ultra and Team. Scheduled tasks run in the cloud, but anything touching local folders needs the desktop app open. (Cowork help, scheduled tasks)

The Highest-Leverage Scheduled Task
Every Monday at 8am: Read the project files in /active-projects and my Notion boards, and check Gmail for anything from clients over the weekend. Create a weekly priorities doc in /briefings named "[YYYY-MM-DD] Weekly Priorities" with: - CLIENTS WAITING: any client message that needs a reply today, flagged at the top - TOP 5: the five things I need to move this week, ranked, with one line on why each matters - AT RISK: deadlines this week with no recent movement Rules: never send email, never delete files. If nothing came in from clients, say so in one line.
Build Your Setup For Your Role

Fill in the brackets and paste this into a new chat. You’ll get Project instructions, your first connectors, a delegated task and an example outcome prompt, all for your job.

Vibe Working Setup Prompt
I want to set up Claude properly for vibe working: delegating outcomes, not managing tasks. Help me build the full setup for my role. My role: [job title and 2 to 3 sentences on what you actually do day to day] My industry: [e.g. marketing, finance, operations, healthcare, legal] Tools I use most: [e.g. Gmail, Notion, Slack, Google Drive, Salesforce] What eats most of my time right now: [2 to 3 things that take too long or happen too often] Step 1: Write my Project instructions Based on my role and context, write a complete Project instruction set I can paste into a new Claude Project right now. Cover who I am and what I do, my communication style and tone, standing rules Claude should always follow in this Project, what "good output" looks like for my work, and anything Claude should always ask before doing. Step 2: Tell me which connectors to set up first Given the tools I listed, which 2 to 3 connectors should I set up this week for maximum impact? For each, give me one specific prompt I could run the moment it's connected that would immediately save me time. If a tool I listed has no Claude connector, say so plainly. Step 3: Write me my first delegated task Based on what's eating most of my time, write a complete task brief I can hand Claude right now: the specific outcome, the file format, which folder or files to use, what to do with edge cases, and a request to show me the plan before executing. Then tell me whether it's worth putting on a schedule and at what cadence. Step 4: Give me my first outcome-delegation prompt Write an example of a well-structured outcome-delegation prompt for one of the tasks I described. It should give Claude full context, a clear outcome, and everything it needs to finish without coming back to ask me questions. Make everything specific to my role. Don't invent tools or features I didn't mention. The goal is to walk away with a setup I can use today.
The Mistakes To Skip

01

Still prompting like it’s a chatbot

Short prompts with no context get short answers. Brief it like an agent. A prompt that takes 3 minutes to write can save you hours.

02

Projects with no documents

Instructions alone are good. Instructions plus real samples of your work are what make output sound like you.

03

Setting up everything at once

Pick one layer. Set up one Project this week and use it for everything in that area, then add the next layer.

04

Never defining done

"Create a 1-page summary as a Word doc in /outputs" gets you something usable. "Summarize this" gets you whatever Claude decides a summary is.

05

Delegating the wrong things

Relationship calls and high-stakes judgment stay yours. When in doubt, hand off the prep work and keep the decision.