OpenClaw Use Cases That'll Make You Rethink What AI Agents Can Do

These OpenClaw use cases go from dream agents to real automations. See what OpenClaw is, what people are building, and whether it is for you. Then vote on what I build next for you.

The time to talk about OpenClaw has finally come.

I didn't jump on it when it first dropped. The early use cases I kept seeing online just weren't inspiring me.

Organizing your daily to-do list. Retrieving emails. And other things you can already do with tools that are far easier to set up, like Claude Cowork (I shared here how to get automatic morning briefings from your calendar and email), n8n automations like my personal assistant, or even Gemini.

Unless you're a business drowning in inbound and juggling a hundred moving pieces, stick with basic setups. If the job can be done with simpler tools, it should be.

That's partly why I didn't jump in immediately when it went viral. But then I started getting my own ideas on what I'd actually use an agent like this for.

Things I’ve always wanted to do but never had the time for. Things that would actually grow my newsletter and tap into the real power of how work is about to change.

So I set it up. And for the past 2 weeks, I’ve been building with it.

It’s been a lot of work. Making it secure, training it, giving it access to everything it needs, configuring its memory, setting up skills. None of that is trivial.

Which is why I’m thinking about setting this up for you. Building agents tailored to what you need, so you can skip the technical headache and go straight to the power.

But to do that, I need to know what would matter to you. So today I’m sharing ideas for what an OpenClaw agent can do. Some from my own setup, some I picked specifically because I know what you’re building.

Read through them, and then tell me, whether through a vote, a comment, or a reply to this email, what would help you most.

---

Here’s what we’ll cover

• What OpenClaw is (a quick overview)

  • The difference between an AI tool and an AI agent

  • How OpenClaw works under the hood

• Why OpenClaw is different from everything else and what the hype is about

• My own OpenClaw agent and what I’ve built with it so far

• 3 real OpenClaw agents on a mission (sharing their journey publicly)

• 5 OpenClaw use cases: what it could do for your work

• Vote for the use case that fits you best

• Where this is going

Already know what OpenClaw is? Jump straight to the use cases.

What is OpenClaw?

OpenClaw is an open-source AI agent. It was created by Peter Steinberger, an Austrian developer, as a side project back in November 2025. It went through a few name changes (Clawdbot, then Moltbot, then OpenClaw), exploded in popularity in late January (60,000 GitHub stars in 72 hours, now sitting at nearly 240,000), and as of mid-February, Steinberger announced he's joining OpenAI while the project moves to an open-source foundation. Things moved fast.

But none of that matters as much as what it does. And to understand that, you need to understand one thing first.

The difference between an AI tool and an AI agent

• Regular LLMs (ChatGPT, Claude, Gemini): You write a prompt. You get an output. Then you do something with that output. You copy it, paste it, edit it, send it, upload it. The AI thinks. You act.

• AI agents like OpenClaw: You don’t give it a prompt. You give it a job. And it doesn’t wait for you to babysit between steps. It just does it.

Here’s the kind of job you’d give it: “Whenever a new lead fills out my contact form, research their company, find their LinkedIn, figure out what they actually need based on their industry, pull the three most relevant case studies from my portfolio, build a personalized proposal, attach it to an email that sounds like I wrote it, send it within 5 minutes, log everything in my CRM, ping me on Slack with a summary, and if they don’t reply in 3 days, follow up.”

Form. LinkedIn. Website. Portfolio. Proposal. Email. CRM. Slack. Follow-up. Ten apps. One job. Zero you.

That’s what “autonomous agent” means. Not a chatbot that talks well. Software that acts on your behalf, end to end, without you sitting there the whole time.

How OpenClaw works under the hood

• It runs on your own machine. Your Mac Mini, a server. It connects to an LLM of your choice through the API (Claude, GPT, Gemini), and that model becomes its brain. The thinking happens in the cloud. The doing happens locally, on your hardware. Your data stays with you.

• You talk to it through apps you already use. WhatsApp, Telegram, Slack, Discord, Signal. You send it a message the same way you’d text a person. And it doesn’t just reply. It executes. Browse the web, organize your files, write code, even control a browser to log into sites and click around on your behalf.

• It remembers everything across sessions, so it gets sharper the longer you use it. And you’re not limited to one assistant: you can create a “Work Agent” for Slack and a “Home Agent” for WhatsApp, keeping their brains entirely separate.

• It’s endlessly expandable. Skills let you download new capabilities from the community or ask the agent to build one for you on the spot. Plugins give it physical abilities, like dialing a phone number to make a voice call. And for bigger projects, it can spawn sub-agents (think digital interns) to handle background tasks while your main assistant stays free.

But all of this power comes with real complexity. Setting it up requires the command line. Securing it means understanding what you're exposing. And if you don't know what you're doing, the risks are real. (If you want to go deep on the security side, ToxSec has been covering this thoroughly.)

So why is everyone losing their minds over it?

---

What makes OpenClaw different

Here’s how OpenClaw compares to the tools you probably already use:

That last row is important. But the row that has everyone excited is “when it works”. Because this is what it actually looks like:

You tell it once: “Every weekday morning before the market opens, analyze these 10 stocks I’m following, check if my portfolio is overweight in any single industry, compare performance against my target allocation, and if anything’s off by more than 5%, rebalance automatically. Then send me a summary of what you did and why so I can review it over coffee.”

You didn’t open an app. You didn’t check a dashboard. You didn’t log into your brokerage. You woke up and it was already done.

How it does that

Because OpenClaw runs on a dedicated machine, it can stay on whether your laptop is open or not. You can shut your computer, go to bed, fly to another country. It’s still working.

And it uses two systems to act on its own:

• Cron jobs are scheduled tasks you set once. “Every night at 1am while I’m sleeping, go through today’s new Twitter followers, find the ones who have a business in their bio, look them up on LinkedIn, write a personalized DM based on what they do, and send it before I wake up.” It runs on a timer, whether you remember to ask or not. (Similar to the scheduled tasks in Claude’s Cowork, which I covered here.)

• Heartbeats are different. Every 30 minutes, OpenClaw wakes up on its own, runs through a checklist you’ve defined, and decides if anything needs your attention right now, using the full context of your conversation history.

This is why people are losing their minds over OpenClaw. It’s a tool that works while you don’t.

---

My OpenClaw agent setup: what I built in 2 weeks

Meet Otto. He's my OpenClaw agent, and his official role is "Head of Everything Daria Doesn't Want to Do” (but should, if she wants to grow the newsletter😅).

You can find him on X. He posts as himself, not as me. He has a dry sense of humor and he’s already getting annoyed when I assign him admin work instead of tasks that actually help grow AI blew my mind. (Yes, it’s wild to talk about an AI agent like this. I know.)

The first week was all setup and small jobs: getting it running, configured, secured, writing its files, defining its memory structure, setting its rules. And then research tasks, admin work, filling out spreadsheets, pulling data. I started small on purpose, because I wanted to understand how it works before handing it anything important.

The real goal: Otto becomes an extension of me. Not a tool I use, but a version of me that handles everything I know I should be doing but will never have time for. Managing X. Posting on Reddit. Adding resources to the LAB. Opening an Instagram and Threads account and keeping them active. All the growth work that matters but always loses to the next deadline. One agent, dedicated to growing the newsletter while I focus on writing it.

It’s still very much a work in progress. But even at this stage, Otto is already taking real work off my plate. (Once he’s further along, I’ll write a dedicated post about the full setup.)

---

3 OpenClaw agents worth following

There are hundreds of these out there right now, and more launching every day. Here are 3 that I’ve been following, so you can see just how wide the possibilities are.

Felix Craft — built by Nat Eliason

Role: Autonomous entrepreneur

What he’s done: Nat gave him 1,000 dollars and said “I’m going to sleep. Build a product that makes money”. And he did. Now, after three weeks, Felix built and sold an info product, launched a marketplace for OpenClaw agents (Claw Mart), earned trading fees from a community-launched crypto token (FELIX), and pulled in over $62,000 in combined revenue. He now pays for his own API costs from what he earns.

Kelly Claude — built by Austen Allred

Role: Product builder

What she’s done: Builds 12+ complete iOS apps per day, designs them, submits them to the App Store, and is now warming up TikTok and Instagram accounts to market them herself. Austen built a custom UI where each column is a separate Kelly session working on different projects simultaneously.

Atlas Forge — built by Jonny Miller

Role: Generative artist

What he’s done: Two models working blind. Codex writes the visual algorithm. Claude Opus evaluates the output against aesthetic principles Atlas developed himself. Neither sees what the other does. He calls this the “Picasso Loop”. 673 renders over 3 days on one series. No AI image generators. He writes code that generates art: flow fields, vortex spirals, warp effects.

---

OpenClaw use cases: 5 ways to put it to work

These use cases are meant as inspiration, based on what I know about you, the people reading this newsletter, so you can see how far this could go.

The solo consultant who sells time

The problem: You’re great at what you do. But between delivering work, you never get to the stuff that brings in more work. Your LinkedIn’s been dead for weeks. Your website still shows last year’s offer. You have case studies sitting in Google Docs that never made it anywhere. There’s always a deadline. There’s never time for marketing.

What OpenClaw does: It keeps your social platforms active, updates your website with new offers, publishes those case studies, drafts follow-ups for leads going cold, and sends you a daily brief on what needs your attention. You wake up and the marketing you never do is already done.

The side hustler builder

The problem: You’ve had the idea for months. Maybe years. The problem isn’t motivation. It’s everything else. You don’t know where to start or how to do it. You don’t have time to figure out landing pages, payment setups, social media strategies. You come home drained and the idea stays in your head for another week.

What OpenClaw does: It researches your market, builds the product, then a product page, sets up the sales flow, handles customer interactions, and markets it on social media while you sleep. Remember Felix? That’s exactly what he does. You don’t need to quit anything. You need an agent that works the hours you can’t.

The company that wants to scale without hiring

The problem: You’re growing, but there’s more work than people to do it. Leads go cold because follow-ups are late. Campaigns launch half-finished. Everyone’s stretched thin and the bottleneck is always the same: not enough hands. Nobody has budget to hire.

What OpenClaw does: You go to sleep and wake up to 20 demos booked in your calendar. You just have to show up. Admin work is processed before anyone walks in. Campaigns are running on schedule. It works as a shared agent across the whole company, with different access levels for different roles, so everyone gets help where they need it.

The small company that never has enough hands

The problem: NGOs, kindergartens, clinics, family businesses. Information comes from everywhere, phone calls, emails, WhatsApp groups, paper forms, and nothing is centralized. Nobody’s job title includes “keep track of everything”, but someone always ends up doing it anyway, on top of their actual work.

What OpenClaw does: A supplier sends an invoice by email. It gets filed, matched to the right budget line, and added to the payment queue. Someone asks a question in the WhatsApp group. The answer gets pulled from internal docs and posted automatically. Multiply that across every admin task nobody wants to own.

The content creator drowning in distribution

The problem: You spend hours on a piece. You hit publish. And then it just... sits there. Thirty views. Not because it’s bad, but because nobody saw it. The repurposing, the scheduling, the cross-posting, the engaging in comments... that’s another full-time job. And you never wanted that job. You wanted to create.

What OpenClaw does: It takes your published piece, repurposes it across platforms, creates images and clips, posts and engages with comments, tracks what’s performing, and suggests what to write next. You do what you love. It does what you’d never get around to. Your work finally gets seen.

---

Vote on the one that fits you best

If none of them do, reply to this email or drop a comment and tell me what you’d need in your own work.

---

Where this is going

Every single day, agentic AI takes another step forward. We see it with everything Anthropic is launching with Claude Cowork and Claude Code. We see it with Cursor. With Google’s products. With Perplexity, which just launched Perplexity Computer. It’s everywhere. This isn’t the future of work. It’s already the present.

And the thing we all need to keep practicing is expanding what we think is possible. The limits you assume exist today might not exist tomorrow. That sounds dramatic, but it’s just how fast things are moving right now.

Is OpenClaw for everyone? No. If you’re thinking about buying a Mac Mini just to sort your personal inbox or handle your to-do list, it’s not worth it. The tools you already have access to can do that, and they’re far easier to set up.

But if you’re a business where the volume justifies it, or if you’re tired of all the work you have to do but hate doing, then this is worth paying attention to.

And there’s another reason to start now: the people who learn how to work with agents early are the ones who’ll be ahead when this becomes the default.

Tell me your use case. And I’ll be thinking, on my side, about ways I could help you get there.

And if you found this useful, share it. I’d love for this to reach people who are still only seeing the basic use cases and haven’t realized how much further this can go. The more of you who tell me what you need, the better I can figure out how to help.

---

This post is free. If you found it useful and want access to more of what I’m building - prompts, automations, step-by-step guides - paid subscribers get all of it.