This is Co. · your free guide

What should you
actually automate?

{the un-techy way to get Claude working while you're offline}

🎧

me, in your ears

ps - this whole thing is an episode

Want to hear me talk it through?

This guide came straight out of Episode 56 of the pod - the one about the Claude update that means your biz can keep running while you're offline. If you want the full behind-the-scenes of the automations I actually use, press play and read along 🎧

Apple Podcasts Spotify Other players

First, the honest bit

Automation is sexy. Automating the wrong thing isn't.

Here's the thing nobody says out loud when they're selling you on agents and automation: I do not believe everything in your biz should run end to end without you. That's not the goal. It makes complete sense for us humans to stay in the loop for so many things, even while AI takes the load off. There are also things you can happily be OUT of the loop for - stuff Claude can just do. The whole game is knowing which is which.

'Cause here's what goes wrong. People rush off to automate before the thing even works - and all you get handed back is a pile of stuff you hate. So before we talk about setting anything on autopilot, the real question isn't "what CAN AI do?" It's this: what repetitive work could I hand over, so I get to focus on the parts of my biz only I can do? That's where the good stuff lives.

This little guide walks you through exactly how I decide - and how I build it so it actually works instead of handing me back junk. Four steps. No tech degree required. Let's go 💛

Step one

Start with your constraint.

Don't automate everything. Automate your bottleneck.

The mistake I see is people trying to build everything at once - a shiny agent for every corner of the biz. Please don't. When I started automating, I did my constraints finder first. Every month I check in with myself: what's the real bottleneck right now - is it capacity, is it revenue, is it offer delivery? Then I train up help in that ONE area. Find your biggest constraint and start there. At one point mine was time, plain and simple - so that's exactly what I went after first.

Fill in the blanks - do it now

Right now, my biggest constraint is [revenue / offer delivery / capacity + time].
The task inside that constraint that eats the most of me is [the recurring thing you dread].
It steals about [X hours] from me every [week / month].
So THAT is the one I'm looking at first - not the ten other shiny ideas.

Where to use it: a quick monthly check-in with yourself before you build a single thing - it stops you chasing shiny objects instead of your actual bottleneck

Find yours

If sales are quiet, don't go building an agent for your morning routine. What's the one constraint that, if it eased, would free you up the most right now?

Step two

Sort what to hand over, and what to keep.

The in-the-loop line

Once you've got your constraint, break that job into its steps and run each one past a simple question: does this need ME, or can Claude just do it? Take my club deliverables. I love ideating the goodies and making the deliverables - that's mine, that's the fun part, I'm keeping it. But the stuff in between - making the Canva tile, uploading it, posting it, popping the to-dos on my calendar - I don't need to be the one doing that. So THAT'S what goes over. You draw the line on purpose, and the golden rule stays: nothing goes out to my audience without me seeing it first. That part never changes.

Sort it - two columns

Hand over {recurring, rules-based, back-of-house}:
[e.g. pulling the report, making the tile, retrieving the transcript, drafting the first pass]

Keep {the soul, the eyes, the ideas, anything people-facing}:
[e.g. the idea, the final read, the tweak, the yes-or-no, the strategy call]

How this plays out in my biz

My podcast-to-freebie flow: an agent grabs the transcript, drafts the freebie, spins up the opt-in copy - I take back the last bit and add the soul in. My club admin: I still ideate and tick yes; it makes the tile and posts it. My Meta ads: it drafts new angles and hands them over; I test them myself and it never touches a dollar. My chief of staff: I just riff voice notes all day; it sorts them at night. Handed over the plumbing, kept every drop of the soul.

Where to use it: any time a task feels automatable - split it into "needs me" vs "doesn't" before you build, so you never automate the part that's actually your magic

Find yours

Take that constraint task from step one and list its steps. Which ones genuinely need your eyes or your soul - and which are pure plumbing you'd happily never touch again?

Step three

Build it right before you automate.

Automation is the LAST step, not the first

This was never about rushing off and suddenly having some slick automation. It was about breaking my process down first, and building the help first. The magic isn't the schedule - it's the work you do before the schedule. Here's the actual order I follow every single time, and it's the bit most people skip 'cause they run straight to "set it and forget it":

The order - tick them off as you go

1. Make sure Claude can already do the task really well with you still in the loop - if it can't do it well by hand, it won't do it well on autopilot.
2. Build your brain - give Claude the real context about you, your biz, your ideal client.
3. Build the skill - teach it this one task properly, so it acts as YOU, not a random generalist.
4. Connect it - into the tools it needs {Claude Connectors, or a custom connection}.
5. Use it and improve it by hand - prompt it yourself, and every time it comes back a bit off, go fix the skill.
6. Only once you genuinely love what it hands back - THEN set it on a schedule.

How this plays out in my biz

Take my opt-in page writer. You could go to Claude right now and say "make me an opt-in page" - and you'd have no idea if it's any good, if it converts, if it even works on a cold audience. It's a generalist. So instead I bottled up everything I learned growing my list to tens of thousands - what actually converts - into one skill. Now it's not guessing. It hands me back AI that sounds like me, not generic slop. That's the whole difference.

Where to use it: every single automation, in this order - the schedule is the reward at the end, never the starting point

Find yours

For your one task: can Claude already nail it with you in the loop? If not, you're not ready to automate it yet - you're ready to build the skill. Start at step one.

Step four

Keep the soul.

You are the gatekeeper. Always.

Even on the stuff I've automated, there's a rhythm I never break. I give the first 10% - the idea, the spark, the soul. Claude does the middle 80% as me, 'cause it's trained on my voice and my way. Then I take back the last 10% - I read it, I humanise it, I add my little sprinkle. When my podcast freebie lands in my inbox, honestly it usually comes back 90, 95% right and it could go out as-is. But I will not let the robots rob me of that soul sprinkle. It makes me feel too good. And nothing reaches my people that didn't pass through my eyes and my standards.

Your last-10% checklist

Before anything an agent made goes anywhere, I ask:
- Is this [actually in my words - would I say it out loud?]
- Did I add the [one story, line or detail only I have]?
- Would I be proud to put my name on it, exactly as-is?
If yes to all three - it goes. If not - I sprinkle, then it goes.

Where to use it: the final gate on everything that touches your audience - the more you automate, the more this last 10% matters

Find yours

What's YOUR soul sprinkle - the thing that makes a piece unmistakably yours? Name it, so your agents leave room for it and you never skip it.

💛

come build with me

the shortcut, if you want it

You don't have to build these from scratch.

Everything I just walked you through - the opt-in writer, the lead magnet magician, the welcome-sequence writer, the whole team - I've already trained up on marketing tasks and handed over inside the Content Mastery Club. My members install them in a click, tweak them to sound like them, and automate from there. The hard part's done. You just make them yours.

See if there's a spot in the Club

Before you go build it

Take ONE task off your plate. Just one.

Not ten. Find your biggest constraint, pick the one task inside it you'd love off your plate, teach Claude to do it as you, and only automate once you genuinely love what it hands back. That's the whole thing - and it's how your biz starts running while you're off living your life 💛

Carinda xThis is Co.

Download as PDF

this is co.

thisisco.com.au · @carinda.gosling.co · hello@thisisco.com.au

What should you
actually automate?

Want to hear me talk it through?

Automation is sexy. Automating the wrong thing isn't.

Start with your constraint.

Don't automate everything. Automate your bottleneck.

Sort what to hand over, and what to keep.

The in-the-loop line

Build it right before you automate.

Automation is the LAST step, not the first

Keep the soul.

You are the gatekeeper. Always.

You don't have to build these from scratch.

Take ONE task off your plate. Just one.

Take your pick

The fine print

What should youactually automate?

Want to hear me talk it through?

Automation is sexy. Automating the wrong thing isn't.

Start with your constraint.

Don't automate everything. Automate your bottleneck.

Sort what to hand over, and what to keep.

The in-the-loop line

Build it right before you automate.

Automation is the LAST step, not the first

Keep the soul.

You are the gatekeeper. Always.

You don't have to build these from scratch.

Take ONE task off your plate. Just one.

Take your pick

The fine print

What should you
actually automate?