Member access

Member login required

Enter your email and password on the login page to open the members hub.

Use the same email you paid with when you log in or create your member account.

Which platform is your
OpenClaw setup on?

Choose your platform and we'll tailor the local install guide for you.

🪟

Windows PC

Windows 10 or 11
Any modern PC

Recommended
🍎

Mac

M1, M2, M3 or M4
Mini · Air · Pro

Your AI operating system

Welcome back.

Know what to use.
Know what to do next.

Your guides, saved stack, connector recipes and the AI updates that actually affect your work — organised into one next move.

36practical tool guides
12connector recipes
Liveofficial AI updates
0items saved for you
Your stack

Tools you actually use

Add your tools and the hub will filter out irrelevant updates.

What changed

Relevant to your stack

Checking official AI sources…
Your learning path

Small lessons. Real output.

Suggested connection

ChatGPT → Google Drive

Native

Bring your own documents into everyday research, drafting and analysis without repeatedly copying context.

Personalise your dashboardAbout 60 seconds. Saved only on this device.
What do you want AI to improve?
Which tools are already in your stack?
Your choices stay in this browser. No personal workflow data is sent to the news service.
Learning studio

Learn one thing.
Use it today.

Short lessons built around an outcome, a task, a reusable prompt and a quality check. Mark each lesson complete and the dashboard remembers where you left off.

Want the original deep-dive material?The full foundations, prompt library and first-30-minute plan are still here.
Connector atlas

See how your tools
work together.

Search by tool or outcome. Every card explains the connection method, what it is genuinely good for, the difficulty and where to start.

Connection methods and plan availability change. Recipes link to the relevant guide and include a “last checked” date so you know when to verify the vendor’s current plan.

Workflow library

Turn separate tools
into repeatable work.

Pick the result you need, follow the simple chain, and save it when it earns a place in your routine. No automation required for the first run.

Advanced systemsReady for OpenClaw, command recipes and connected agent workflows?
AI signal briefing

The changes worth knowing.
The noise left out.

Fresh posts from official AI-company sources, translated into who should care and whether to act, wait or ignore. Your saved stack is prioritised first.

Refreshing official sources…
Your saved room

Keep what matters.
Skip the rest.

Your chosen stack, bookmarked connections, completed lessons and saved workflows live here. This first version saves privately in this browser.

Stack

Your tools

Connector atlas

Bookmarked recipes

Learning

Completed lessons

Workflows

Saved routines

Private by default. These preferences are stored on this device, not in your member account. Account syncing can be added later without changing your current access.
Members hub
Welcome back.

Choose the right AI tool,
then use it properly.

Start here when AI feels noisy. Pick an outcome, follow a path, run one workflow on real work, then decide what tool belongs in your stack. The hub is organised around jobs to be done: writing, research, business admin, content, coding, automation, and AI agents.

Find my tool See AI news
36member guide pages
7workflow lanes to choose from
1first workflow to run today

Inside your membership

Updated weekly
Ready now
Start HereChoose your first stack, run one workflow, and avoid paying for overlapping tools too early.
Outcome lanesWriting, research, business/admin, content, coding/building, automation, and AI agents.
Guide vaultChatGPT, Claude, Gemini, Perplexity, NotebookLM, Cursor, Midjourney, Runway, ElevenLabs, Canva, Gamma, Zapier, Make, n8n, OpenClaw, and more.
Agent systemsOpenClaw setup, command library, WhatsApp/Telegram workflows, safety checks, integrations, and troubleshooting.
Fresh this week Live product updates

Tool picks, guide wording, setup paths, news links, and the local agent lane are reviewed against what changed this week.

Best next move Browse tool guides

Start with the tile rail below, then jump into the guide that matches the work you want help with first.

Start here Start with the job, not the tool.

Tell the hub what outcome you want and it will point you to the cleanest first tool, plus the next tool to pair with it.

Updated guides Not static one-pagers.

Guides are refreshed around product updates, model launches, creator trends, and the workflows members actually need.

Today Last research pass: 29 June 2026.

News cards link to official company updates or trusted AI news feeds, then translate the update into the guide worth opening.

First 5 minutes

Do this first,
before browsing the guides.

The hub works best when you do not treat it like a library. Pick one job, get one recommended stack, copy one prompt, and run it on real work. If you get a useful output, then open the deeper guide.

1
Choose one job due this week.

Email, research, content, study, admin, coding, automation, or an agent setup. Not "learn AI". A real job.

2
Ask the Stack Finder.

Type the job in plain English. It will recommend the first tool, the second tool only if needed, and what to ignore.

3
Open one guide only.

Use the recommended guide. Run the first workflow before you open another tool or compare more options.

4
Save the reusable version.

Keep the prompt, the output checklist, and the next action. That becomes your first personal AI workflow.

Continue your path First 5 minutes: pick one job, run one workflow, save one reusable prompt.

Start here first. It stops the hub turning into another pile of interesting tools.

Open quick start
Recommended next guide Start with ChatGPT, then choose a specialist only when the job needs it.

Most members should prove one workflow before adding Claude, Perplexity, Zapier, or agents.

Open ChatGPT guide
Search the hub Find a guide or prompt by outcome.
Search jumps to the Tool Finder and pre-filters the Guide Vault below.

Choose your path.

You do not need to learn every tool at once. Start with the outcome you want first, get one real win, then go deeper from there.

Best first 30 minutes

How to get value fast.

Use the hub in this order so you leave with one useful workflow, not a bigger list of tools to feel behind on.

1. Pick one job to improve

Choose writing, research, coding, content, or automation based on the thing you actually need finished this week.

2. Run one workflow today

Open one guide, copy the first workflow, and use it on real work. Fake examples do not teach you where the tool fits.

3. Then build your stack

Add a second tool only when it solves a new job better than the first one. More apps is not the same as more leverage.

Start Here operating plan

Open one path, run one workflow,
then add the next tool only when needed.

This is the paid-hub promise in action. Each path gives you the first guide to open, the first workflow to run, and the next move when the first tool starts to hit a limit.

Use this first
Beginner Get one useful AI win today.

Use when AI feels noisy and you do not know where to start.

  1. Open ChatGPT.
  2. Run the rough-notes brief prompt.
  3. Add Claude or Perplexity only after the first task works.
Open guideLoad stack
Business owner Turn messy admin into business assets.

Use for client replies, briefs, SOPs, decks, diagrams, and weekly operations.

  1. Open ChatGPT or Copilot.
  2. Create one SOP, reply, or brief.
  3. Move repeat tasks into Zapier or Make.
Open guideLoad stack
Creator Build a repeatable content machine.

Use when you need hooks, scripts, captions, visuals, clips, and repurposing.

  1. Plan with Claude or ChatGPT.
  2. Create assets in Canva or CapCut.
  3. Repurpose with OpusClip or Runway.
Open guideLoad stack
Student Study from sources, not shortcuts.

Use when you need summaries, quizzes, explainers, and source-grounded understanding.

  1. Research with Perplexity.
  2. Load docs into NotebookLM.
  3. Use Claude to explain gaps.
Open guideLoad stack
Builder Prototype fast, then review properly.

Use when you want a website, app, code fix, dashboard, or working prototype.

  1. Prototype in Lovable.
  2. Edit and inspect in Cursor.
  3. Use Claude Code only when you can review diffs.
Open guideLoad stack
Automation Remove one repeat task safely.

Use when the same admin, lead, report, or routing task happens every week.

  1. Map the manual process.
  2. Build the smallest Zap or Make scenario.
  3. Add n8n when control matters.
Open guideLoad stack
AI agents Move from prompts into supervised systems.

Use only after one workflow is proven and worth keeping always-on.

  1. Define trigger, tools, and forbidden actions.
  2. Install OpenClaw and connect one channel.
  3. Run read-only before adding write access.
Open guideLoad stack
First 30 minutes

Use the hub once,
and leave with one working AI workflow.

This is the fastest way to make the membership feel useful. Do not browse all 36 guides first. Pick one real job, run one workflow, save the result, then decide what tool you need next.

Member rule: one finished workflow beats ten interesting tabs. This section is designed to get a beginner, business owner, creator, student, builder, or automation user moving immediately.
1
Pick one real task

Use something due this week: an email, customer reply, study note, post, competitor check, workflow map, or small build.

2
Choose the closest path

Open Beginner, Business, Creator, Student, Builder, Automation, or AI Agents. The path tells you what to ignore.

3
Open the first guide only

Run the first workflow before opening another tool. If ChatGPT or Claude can solve it, do not complicate the stack.

4
Copy the starter prompt

Paste your real context, force an output format, and ask for a quality check before accepting the answer.

5
Save the reusable version

Keep the prompt, the checklist, and the improved output. That becomes your first personal AI workflow.

Prompt library

Copy-ready prompts for real work,
not generic prompt tricks.

Use these when you know the job but need a clean starting point. Each prompt forces context, constraints, output, review, and next action so the result is usable.

The full guide pages expand these into tool-specific prompt packs. This dashboard gives members the fastest version first.
Tool chooser Pick the right AI stack

Use when the member knows the problem but not the tool.

I need to [goal]. My situation is [context]. My constraints are [budget, time, skill level]. Recommend the simplest AI stack, the first tool to use, the second tool only if needed, and what to avoid.
Beginner
Business Turn messy admin into an SOP

Use for repeat customer, operations, or internal work.

Convert this messy process into a clear SOP: [paste process]. Include trigger, inputs, steps, owner, approval point, failure case, and what can safely be automated later.
Business
Email Write a better reply

Use when tone and next step matter.

Draft a reply to this email: [paste]. Goal: [goal]. Tone: clear, warm, direct. Include a short answer, next step, boundary if needed, and a final version under [word count].
Claude
Research Create a source-backed brief

Use with Perplexity, NotebookLM, Claude, or ChatGPT with sources.

Research [topic] for [decision]. Give me a source table, key findings, conflicting views, what is probably true, what is uncertain, and the action I should take next.
Research
Creator One idea into a content pack

Use before Canva, CapCut, Runway, or Midjourney.

Turn this idea into a content pack: [idea]. Audience: [audience]. Give hooks, short script, carousel outline, caption, visual direction, and one repurpose plan.
Creator
Automation Find the safest first automation

Use before Zapier, Make, n8n, or OpenClaw.

Audit this workflow for automation: [workflow]. Tell me what should stay manual, what can be automated safely, what needs approval, and the smallest first automation to build.
Automation
Student Learn from sources

Use with notes, PDFs, readings, and exam prep.

Using only these notes/sources, teach me [topic], then quiz me. Include definitions, examples, likely exam questions, and the gaps I need to revise.
Student
Builder Make coding agents safer

Use before asking Cursor, Claude Code, or ChatGPT to change code.

Before editing, inspect this task and propose the smallest safe code change. Tell me files likely affected, tests to run, risks, and what not to touch.
Builder
Agents Design an agent with guardrails

Use before local agents, channels, APIs, and always-on systems.

Design an AI agent for [job]. Define trigger, allowed tools, forbidden actions, approval points, log format, failure handling, and the first read-only test.
Advanced
AI news radar

What changed in AI,
and what it means for you.

Use this as the member briefing layer. The goal is not to read every AI headline; it is to know which changes affect your tool stack, your content, your business, and the guide worth opening next.

Updated 29 June 2026. Each card turns the headline into a practical member move so the hub feels current, not static.
OpenAI25 Jun 2026
Agent usage is moving from demos into daily work.

OpenAI says Codex has become the primary internal AI work tool across departments, including non-technical teams.

Member move: open AI Agents, Claude Code, and n8n. Build one boring workflow with logs and review before chasing autonomous systems.
Anthropic23 Jun 2026
Claude Tag brings Claude into Slack workflows.

Claude can be invited into selected Slack channels, use connected tools and context, and help teams delegate work inside chat.

Member move: open Claude and AI Agents. Notice the pattern: useful agents live where the work already happens.
OpenAI Codex2 Jun 2026
Codex is becoming a knowledge-work tool, not just a coding tool.

OpenAI reports Codex has over 5 million weekly active users, with knowledge workers becoming a fast-growing group.

Member move: open Builder, Claude Code, and business workflows. Treat coding agents as operating support, not just developer tools.
AnthropicJun 2026
Claude Opus 4.8 strengthens agentic and long-context work.

Anthropic highlights better judgment, tool use, and dynamic workflows for bigger Claude Code tasks.

Member move: open Claude and Claude Code when the job needs careful reasoning, review, and reliable multi-step execution.
Runway5 Jun 2026
Runway API adds faster video generation and stronger editing routes.

Runway's changelog points to Seedance 2.0 Fast, Aleph 2.0, and video workflows with keyframes, references, and audio.

Member move: open Runway and content workflows when motion is part of the job. Keep editing and review separate from generation.
Future ToolsDaily feed
Creator-facing AI news is moving too fast for one source.

Future Tools curates fast-moving stories across AI models, media tools, coding agents, creator software, music, video, policy, and business adoption.

Member move: use this as trend signal, then come back here and open the guide tied to the tool or workflow.
OpenAI Anthropic Google AI Future Tools The Rundown The Decoder
Tool finder

Tell the hub what you are trying to do,
then open the right guide.

Most people do not need 36 tools. They need the right first tool, the right second tool, and permission to ignore the rest until later.

This is intentionally opinionated. The recommender points members toward practical stacks, not random AI tool collecting.
Ask the hub

Use normal language. The hub will route you into the closest member path and show the first tools to open.

Try: better emails, research competitors, make reels, automate leads, build a website.
Recommended stack

Beginner AI stack

Start with one daily assistant, one thinking/writing tool, and one source-backed research tool. This gives you range without drowning in apps.

Learning paths

Pick a path,
not a random tool.

These paths turn the guide library into a course-like experience. Start with the path closest to your real outcome, then move through the tools in order.

Path 01Beginner Start Here

ChatGPT, Claude, Perplexity. Learn how to ask better, check answers, and turn one real task into useful work.

Open first guideLoad stack
Path 02Business Owner Workflow

ChatGPT, Copilot, Gamma, Napkin, Zapier. Build briefs, decks, client replies, simple SOPs, and admin workflows.

Open first guideLoad stack
Path 03Creator Content System

Canva, CapCut, OpusClip, Ideogram, Suno. Turn one idea into hooks, visuals, captions, clips, and reuse plans.

Open first guideLoad stack
Path 04Student and Research Path

Perplexity, NotebookLM, Gemini, Claude. Build source-backed briefs, document summaries, quizzes, and decision notes.

Open first guideLoad stack
Path 05Builder Prototype Path

Lovable, Cursor, Claude Code, DeepSeek. Prototype first, then learn how to review, debug, and improve the output safely.

Open first guideLoad stack
Path 06Automation Ops Path

Zapier, Make, n8n. Remove one repeat admin task before trying to build always-on agents.

Open first guideLoad stack
Path 07AI Agent Systems

OpenClaw, n8n, Claude, ChatGPT. Move from prompts into supervised systems with triggers, logs, and approvals.

Open first guideLoad stack
AI foundations

Know the layer you are using,
then pick the right tool.

Most AI confusion comes from treating every product like the same thing. This quick map shows what is actually happening before you move into the guide library.

A customer does not need jargon. They need to know whether they are asking a model, using an app, building an automation, or installing a local system.
01 LLM

The model brain. It predicts and reasons through language, but it does not automatically know your apps, files, calendar, or business process.

02 AI assistant

The chat app around the model. ChatGPT, Claude, and Gemini give you the interface, files, memory, projects, and daily work surface.

03 Automation

The repeatable plumbing. Zapier, Make, and n8n move data between apps, trigger steps, and remove admin when the process is already clear.

04 Agent

A model plus tools, instructions, context, and safety rules. It can work across steps, but it still needs boundaries and supervision.

05 Local install tool

Software on your PC or Mac that connects the agent to channels, APIs, files, browser access, and workflows. This is where OpenClaw lives.

The simple rule

Use ChatGPT or Claude when you need thinking and output. Use n8n, Zapier, or Make when a known process repeats. Use OpenClaw when you want a local, always-on agent layer that can connect to your other tools and be controlled from chat.

Installable tools

Tools that live on your computer
and connect into everything else.

This is separate from the normal app library. These tools are installed locally, then connected to AI models, messaging channels, automations, browser sessions, and business apps.

Start here only after you have one real workflow worth systemising. Local agents are powerful, but they should solve a clear job.
Featured local install

OpenClaw

Install OpenClaw on your PC or Mac, connect WhatsApp or Telegram, add the model/API layer, then plug it into tools like n8n, Notion, HubSpot, Shopify, Slack, Airtable, Gmail, Calendar, and browser-based workflows.

Runs locally Chat controlled API ready Workflow bridge
Open OpenClaw setup
Step 1 Install on PC or Mac

Get the local app running first. This gives the agent a machine to operate from instead of staying trapped in a browser chat window.

Step 2 Connect a channel

Use WhatsApp or Telegram as the control surface so you can send commands, request briefs, and trigger workflows from chat.

Step 3 Add model and tool access

Connect search, browser access, APIs, and business tools carefully. More access is not better unless the workflow needs it.

Step 4 Pair with your stack

Use OpenClaw with ChatGPT or Claude for reasoning, n8n or Make for repeat workflows, and your real apps for useful outcomes.

Readiness test Do not install an agent until the job is clear.

Agents are for repeat workflows, not vague experiments. Start when the manual version already exists.

  • One trigger starts the workflow.
  • One output proves it worked.
  • One person reviews risky actions.
Prompt: Turn this manual task into an agent-ready SOP with trigger, inputs, output, owner, and failure case.
Permission model Read-only first, approvals second, autonomy last.

The fastest way to break trust is giving an agent write access before you understand its failures.

  • Start with summaries and drafts.
  • Ask before sending, deleting, buying, or changing files.
  • Keep logs for every run.
Prompt: Define allowed tools, forbidden actions, approval points, and logs for this agent workflow.
First workflows Pick boring workflows that save time weekly.

The best first agents are not flashy. They make recurring work less manual and easier to review.

  • Daily briefing.
  • Lead follow-up draft.
  • Weekly report summary.
Prompt: Design a daily briefing agent that summarizes priorities and asks before taking action.
Failure rules If it cannot explain the source, it cannot act.

A useful agent should know what it used, what it did, what failed, and what it needs from you.

  • No source, no decision.
  • No confidence, no customer action.
  • No log, no repeat automation.
Prompt: Audit this agent workflow for unsafe actions, missing logs, and unclear fallback steps.
Guide vault

The full 36-guide library,
grouped by outcome.

Use this when you already know the lane you care about. If you are unsure, start with the tool finder above instead of browsing alphabetically.

The vault is grouped by what members are trying to do: assist, research, create, produce, build, and automate.
Assistants ChatGPT Claude Gemini Microsoft Copilot Grok DeepSeek
Research and docs Perplexity NotebookLM Gamma Napkin
Design and visuals Midjourney Adobe Firefly Canva Ideogram Recraft Krea
Video and audio Runway Google Flow Luma Pika CapCut Submagic HeyGen ElevenLabs Suno OpusClip Descript Higgsfield
Build and automate Cursor Claude Code Lovable Manus n8n Zapier Make OpenClaw setup
No guide matched that search. Try a job word like writing, research, video, automation, business, study, code, or agents.
Daily assistant

ChatGPT

Your broad daily assistant for planning, drafting, summaries, files, and general work when you want one strong all-rounder.

Best for

Writing, planning, notes, summaries, and everyday work when you want one strong default tool.

Start here

Use it on one real work task first, then open the detailed guide below when you want a cleaner workflow.

Open guide Local install tools
CHATGPT
Assistants Research Build Creative Automation Local install Trend radar
AI tool library Browse the apps by outcome. Pick a tool, then open the full guide when the fit is clear.
Social trend radar

Tools getting creator attention
on TikTok, Reels, and YouTube.

These are not automatic subscriptions. They are the social tools worth understanding because they keep showing up in short-form, creator, video, music, avatar, and vibe-building workflows.

Use this lane to test trends without bloating your core stack. If the tool does not help you publish, sell, or build faster, skip it.
VideoGoogle Flow / Veo

Prompt-led scenes, YouTube creator tie-ins, and native-audio video experiments.

 CinematicHiggsfield

Social-native camera moves, viral motion presets, and short vertical visuals.

 EditorCapCut

Templates, auto captions, long-to-short clips, and fast Reels/Shorts editing.

 CaptionsSubmagic

Talking-head captions, B-roll, hooks, silence removal, and clip repurposing.

 AvatarHeyGen

Avatar explainers, translated videos, and multilingual social clips.

 AudioSuno

AI songs, hook audio, parody tracks, jingles, and music-led social ideas.

 BuilderLovable

Vibe-coded websites, internal tools, app prototypes, and product demos.

 CodeClaude Code

Agentic coding, terminal workflows, repo tasks, and advanced builder experiments.

Your tool guide library.

Every card below gives you a clean starting point: what the tool is best at, how to begin, and a first workflow you can run right away. You do not need all of them. Start with one, get a win, then add the next piece deliberately.

Open all 36 guide pages
Daily assistant

ChatGPT

Updated GPT-5.5 refresh

Choose a tool from the library above or the guide rail here to see a deeper walkthrough.

Use this when

Set up first

Standout right now

Pair next

Member shortcut

Open full page
Local install tool

Install OpenClaw and turn AI into a connected system.

Once you want more than chat, this is where you move into a local tool that runs on your own machine. Choose your platform, work through the OpenClaw setup, then use the command library, workflows, and troubleshooting sections below as your operating manual. For most members, this becomes the second phase after they understand the basics and want a supervised system that can connect into the tools they use.

Runs on your PC or Mac WhatsApp or Telegram control Connects APIs and workflows
OpenClaw setup Command library Ready-made workflows Troubleshooting
Command centre
Read-only safe Approval required Advanced
claw give me my morning briefing and flag anything urgent
claw draft follow-ups for overdue leads, but do not send them
claw audit this workflow for unsafe actions before automation
TriggerContextDraftApprovalAction log
Action history mockup Human in loop
09:00
Read-only briefing created

Calendar, email subjects, and task list summarized. No messages sent.

09:08
Follow-up drafts prepared

3 drafts generated and held for approval before customer contact.

09:14
Automation blocked

Financial action detected. Manual approval required before any workflow can continue.

Windows PC Guide

Install OpenClaw
on your machine

Follow the setup once, mark steps complete as you go, then keep this area as your operating manual whenever you need to reconnect channels, troubleshoot errors, or expand what your local agent can do.

🦞

YOU'RE LIVE!

Your personal AI agent is now running 24/7.
Copy any of these and send it right now on WhatsApp or Telegram.

First messagehey what can you do?
Morning routineclaw give me my morning briefing
Check calendarclaw what do I have on today
🦞

Setup Assistant

Paste any error — get an instant fix

Hey! I'm your setup assistant. Paste any error message or describe where you're stuck and I'll give you the exact fix. 🦞
Can't type in terminal
WhatsApp logging out
API key error
Command not found
Browser dashboard
Cancelled wizard
Set up Telegram
Brave Search not working
🦞

OpenClaw command center.

This is the live resource bank for the local install lane: tool connections, command prompts, reusable workflows, pro tips, and a full troubleshooting reference. Bookmark it and come back whenever your setup needs a tune-up.

Do This First

Connect Your Tools

Before the commands below will work, your agent needs to know where your data lives. Here's exactly what works out of the box — and how to connect everything else in minutes.

✓ Works Right Now — No Setup Needed

Your agent has browser access on your machine. If you're already logged into these on your PC, it can use them immediately.

📧 Gmail
📅 Google Calendar
📊 Google Sheets
🐙 GitHub
💼 LinkedIn
📸 Instagram
💳 Stripe
🌐 Brave Search
💬 WhatsApp
✈️ Telegram
One-Time API Setup
⚡ Needs a Quick Connection

These tools need an API key so your agent can read your real data — sales figures, leads, databases. Each one takes under 5 minutes. Click to expand the setup steps.

🛒ShopifySales · orders · customers · products+
  1. In your Shopify admin go to Settings → Apps → Develop apps
  2. Click Create an app — name it "OpenClaw"
  3. Under Admin API access scopes enable: read_orders, read_customers, read_products
  4. Click Install app → copy the Admin API access token
  5. Send this to your agent:
claw remember: my Shopify store is [yourstore.myshopify.com] and my Admin API token is [paste token]

After this, commands like "claw give me my weekly sales report" will pull real data from your store.

📈HubSpotCRM · leads · deals · contacts+
  1. In HubSpot go to Settings → Integrations → Private Apps
  2. Click Create a private app — name it "OpenClaw"
  3. Under Scopes add: crm.objects.contacts.read, crm.objects.deals.read, crm.objects.companies.read
  4. Click Create app → copy the access token
  5. Send this to your agent:
claw remember: my HubSpot API token is [paste token]

Your agent can now pull lead counts, deal stages, and contact history on demand.

📝NotionDatabases · notes · tasks · wikis+
  1. Go to notion.so/my-integrations → click New integration
  2. Give it a name, select your workspace → click Submit
  3. Copy the Internal Integration Token
  4. Important: in Notion, open each page or database you want accessible → Share → Invite → select your integration by name
  5. Send this to your agent:
claw remember: my Notion API key is [paste token]

Your agent can now read and update any Notion page or database you shared with the integration.

💼SlackMessages · channels · team notifications+
  1. Go to api.slack.com/apps → Create New App → From scratch
  2. Name it "OpenClaw", select your workspace
  3. Go to OAuth & Permissions → Bot Token Scopes → add: channels:read, chat:write, users:read
  4. Click Install to Workspace → copy the Bot User OAuth Token
  5. Send this to your agent:
claw remember: my Slack bot token is [paste token]

Your agent can now send messages to Slack channels and read channel activity.

📦AirtableDatabases · spreadsheets · project tracking+
  1. Go to airtable.com/create/tokens
  2. Click Create new token — name it "OpenClaw"
  3. Add scopes: data.records:read, schema.bases:read
  4. Select which bases to give access to → click Create token and copy it
  5. Send this to your agent:
claw remember: my Airtable token is [paste token]

Your agent can now query any Airtable base you gave it access to.

🔄n8nAdvanced automation · webhooks · multi-app workflows+
  1. Install n8n locally — open a terminal and run: npx n8n
  2. Open the n8n dashboard at localhost:5678
  3. Create a new workflow with a Webhook trigger node — copy the webhook URL
  4. Build your workflow — connect Shopify, Gmail, Slack, or any apps you want
  5. Tell your agent about each webhook:
claw remember: my n8n webhook for [workflow name] is [paste webhook URL]

n8n is the most powerful option — it lets your agent trigger complex multi-step automations across any combination of tools.

Copy & Send

The Command Library

Every command below is ready to send on WhatsApp or Telegram. Hit copy, paste it in, customise the [brackets] for your situation. Your agent does the rest.

📊Reports & Analytics
claw give me my weekly sales report
claw what was our best-selling product last month
claw compare this month vs last month revenue
claw how many leads came in this week
claw pull our top 5 customers by spend this quarter
claw what is our conversion rate this month
📧Email & Communication
claw summarise my unread emails
claw draft a follow up to [name] about [topic]
claw who hasn't replied to my last email
claw write a cold outreach email for [product or service]
claw flag any emails that need a reply today
claw send the invoice to [name] for [amount]
📅Calendar & Scheduling
claw what do I have on today
claw schedule a call with [name] on [day] at [time]
claw what meetings do I have this week
claw block 9-10am every weekday as deep work time
claw remind me to follow up with [name] tomorrow morning
🌐Research & Web
claw research [topic] and give me a 5 point summary
claw what's the latest news on [topic]
claw find the best [product] under $[price]
claw check what my competitors are saying about [topic]
claw find the contact details for [company name]
📁Files & System
claw open [app name]
claw find the file called [filename]
claw run my weekly backup
claw check if my website is up
claw take a screenshot of my screen
🎨Content & Social
claw write a LinkedIn post about [topic]
claw draft 5 Instagram captions for my photos this week
claw write my weekly newsletter on [topic]
claw create a 7-day content calendar for [brand]
claw repurpose this blog post into 3 social media posts
💰Finance & Tracking
claw how much did I spend this week
claw check my crypto portfolio
claw what subscriptions am I currently paying for
claw track my package from [carrier] order [number]
claw what is my ad spend vs revenue this month
One Message. Multiple Results.

Ready-Made Workflows

These are pre-built routines you can send right now. One message triggers a chain of actions. The agent figures out the rest.

🌅
Morning Briefing
Start every day sharp
claw give me my morning briefing
What you get back:
Today's meetings · Urgent emails · Weather · Top news · Pending tasks
📊
Weekly Report
Know your numbers every Friday
claw generate my weekly report
What you get back:
Revenue · Tasks done · Top wins · What needs attention next week
🎯
Follow-Up Sweep
Never let a lead go cold
claw check who I need to follow up with today and draft the messages
What you get back:
List of overdue contacts · Draft follow-up messages ready to send
📝
Content Machine
A week of content in 10 seconds
claw create this week's content plan for my brand
What you get back:
5 post ideas · 1 newsletter topic · 1 blog outline · Posting schedule
🌙
End of Day Wrap
Close out clean every evening
claw give me my end of day summary and set up tomorrow
What you get back:
What got done · What's still pending · Tomorrow's top 3 priorities
🔍
Deep Research
30-minute job done in seconds
claw research [topic] deeply — key findings, sources, actions
What you get back:
Key findings · Source links · Recommended actions · Executive summary
Level Up

7 Tips from Power Users

These are the habits that separate people who get 10x value out of their agent from the ones who use it once and forget it exists.

1

Give it context about you — once

Send this once: claw remember: I run [your business] in [city], timezone [timezone], I prefer bullet point summaries under 5 points. It'll use this in every future response without you repeating yourself.

2

Chain commands together

You don't need to send one thing at a time. Try: claw summarise my emails then draft replies to anything urgent — it'll do both in sequence. Stack as many actions as you want in one message.

3

Use the browser dashboard

Open http://127.0.0.1:18789 on your PC to see your agent's live logs, status, and settings. Ask first: claw what is my gateway token — you need this to log in.

4

Teach it your writing style

Paste a few examples of your writing and say: claw this is how I write — match this style whenever you draft anything for me. Emails, posts, and proposals will sound exactly like you.

5

Set up recurring tasks

Tell it: claw every Monday at 8am send me last week's stats and what I should focus on this week. It runs automatically — you don't have to remember to ask.

6

Be specific to get better output

Instead of claw write an email, say claw write a 3-sentence follow-up to a client who hasn't paid in 2 weeks — firm but polite, no fluff. The more detail you give, the better the result.

7

Put it to work while you sleep

Before bed, send: claw at 7am tomorrow send me a morning briefing and flag any urgent emails that came in overnight. Wake up with everything already done.

Fix It Fast

Troubleshooting Bible

Every common error with the exact fix. Click any error to expand it.

"openclaw: command not found"+

Fix: Close your terminal completely, reopen it fresh, and try again. PATH changes from the installer only apply to new terminal sessions. If still not found, re-run the install command from Step 2.

"ANTHROPIC_MODEL_ALIASES error"+

Fix: Your config is out of date. Run openclaw onboard in a new terminal — this resets your config without losing your settings. Have your API key ready to paste.

WhatsApp QR code expired before I could scan it+

Fix: QR codes expire after 60 seconds. Run openclaw channels login to generate a fresh one. Have WhatsApp open on your phone and ready to scan before you run the command.

WhatsApp says "linked device logged out"+

Fix: WhatsApp Web sessions expire if your phone loses internet or the session times out — this is normal. Run openclaw channels login and scan the new QR code to reconnect.

Agent stopped responding — no reply on WhatsApp+

Fix: The gateway or TUI window was closed. Open a new terminal and run openclaw to restart everything. Important: never close the gateway window — it's the background process keeping your agent alive.

"Invalid API key" or "Authentication failed"+

Fix: Your key wasn't pasted correctly — there's often a hidden space or line break. Go to console.anthropic.com, copy your key fresh, then run openclaw onboard and paste again. Windows users: right-click to paste, not Ctrl+V.

API costs are higher than expected+

Fix: Set a spending cap immediately — go to console.anthropic.com → Settings → Limits and set a monthly limit ($10–15 is plenty to start). High costs usually mean the agent is stuck in a loop. Restart with openclaw and it'll clear.

Browser dashboard won't load at 127.0.0.1:18789+

Fix: The gateway must be running for the dashboard to work. Make sure OpenClaw is running (both windows open). Then ask your agent: claw what is my gateway token — you need this token to log in to the dashboard.

Agent gives wrong times or wrong timezone+

Fix: Tell it once: claw remember my timezone is [e.g. AEST / GMT+10 / EST]. It'll use the correct timezone for all scheduling, reminders, and time references going forward.

Brave Search not working / "search unavailable"+

Fix: Your Brave Search API key may have expired or run out of credits. Go to api.search.brave.com, check your key is active and has credits. Then run openclaw onboard to re-enter it in the config.

🤖
Still stuck on something not listed here? Paste the exact error into the chat assistant in the top-right corner of this page — it'll give you an instant fix.