Doing every job yourself? How to hand off the first one to an AI agent

When there's more work than hands, hand off one job first: the one you repeat most. Write down how you do it, give it to an AI agent, and check every piece it makes for the first week. The quality stays yours, and you get those hours back. The quickest way to get the agent is HostNest, set up for you in about 10 minutes; or use the steps below with an agent of your own.

Why does handing off feel so hard?

Because you're the one who knows how it's done. A vet building software alone put it this way: "realised I was 12 departments in one person". Later in the same post: "I am essentially the bottleneck." Hiring isn't in the budget yet, and a stranger wouldn't do it your way anyway. So the work waits for you.

An agent doesn't fix that by guessing. It fixes it when you tell it your way once, and it keeps it.

Don't have an AI agent yet?

The quickest way is HostNest, a team of Claude agents. You sign up yourself, and the rest comes ready: your own server in the EU (a computer that's always on), the agent, and its wiring. It takes about 10 minutes, most of it the server being built:

1. Account: your email (we send you a 6-digit code) or Google.
2. Free trial: 7 days free.
3. Telegram: a few taps create your bot. The bot is yours.
4. Claude: while your server is built, you read four safety rules. Then you sign in to your paid Claude plan (Pro, Max, Ultra or Team) and paste back one code.
5. Live: open your agent in Telegram and send it the message below.

After that, the upkeep is handled for you. Updates install themselves overnight, a failed one rolls back on its own, and the server is backed up daily. Each agent remembers what you teach it, in notes on its own server. Get your agent on HostNest

In our founder's words: "We're two people. On Monday my agents submitted four listings, put a guide page live, filed the problems owners had posted that day and rendered a film. My part was saying yes and posting."

Do I need to code? No. You talk to your agents in Telegram. How long does it take? About 10 minutes, most of it the server being built. Will it do things without asking? It checks with you before anything goes out or costs money, unless you tell it it doesn't need to. Is my stuff safe? It runs on your own private server in the EU, with your own Claude login, and it's backed up daily.

Already running an agent of your own? The steps below work on any agent that keeps notes between chats. If you write its rules in a file, one job in five lines shows the same idea.

Which job do I hand off first?

The one you repeat most. Look at last week and pick the job that:

• came back again and again (the same kind of email, post, quote or report);
• took hours you'd rather spend on the work only you can do;
• is easy to check: you can tell good from bad in a minute.

Keep the jobs that need you as a person, like a hard talk with a client, for later or for good. Need ideas? Five good first jobs are listed in our guide to agents that keep running.

How do I write down how I do it?

You don't need a manual. Five things are enough:

1. The job, in one line.
2. Your steps, in the order you do them.
3. Two examples done well, pasted in. Your own are best.
4. What must never happen: a send, a promise, a discount, a date.
5. What to show you before anything goes out.

What message do I send my agent?

Copy this, fill in the brackets, and send it to your agent:

I'm handing you one job: [the job, in one line].
Here's how I do it: [your steps, in order].
Here are two examples of it done well: [paste them].
Never: [what must never happen].
For the first week, show me every piece before anything goes out, and tell me what you were unsure about.
When I change something, keep it in your notes and do it that way next time.
After a week, we'll decide together what you can do without asking.

What does the first week look like?

You get each piece in Telegram before it goes anywhere. You say yes, or you say what to change. Each change goes into its notes, so you make it once, not every time.

Ask for proof, not just "done": the draft itself, the link, or the number. How to make an AI agent prove a task is done shows what to ask for. Anthropic's own guide puts the risk plainly: when an agent has nothing to check its work against, "'looks done' is all it has to go on. In practice that makes you the test suite". In week one, you are. That's how the quality stays yours.

At the end of the week, look back. If you barely changed anything, let it do more without asking. If you changed a lot, keep checking.

What won't the agent do?

• It won't send, post or pay without you. It checks with you before anything goes out or costs money, unless you tell it it doesn't need to.
• It won't know your way until you tell it. The first week takes corrections.
• It won't replace the parts of the work that need you as a person.

When one job runs well, hand off the next. How to split work between AI agents covers the second agent.