I Automated My Instagram Carousels and I Haven't Opened the App to Post Since

Instagram turns sixteen next month.

I've been working online since 2008, which means I've been doing this longer than the app has been alive. That's either a credential or a warning depending on how you look at it. (I wasn't there at the start, for the record. It launched in October of 2010, and I showed up a few years later, when I felt like testing it out.)

And somewhere in those years, the part of me that found social media fun quietly packed a bag and left without telling anybody. Posting, engaging, showing up in the comments, feeding the algorithm its little snacks... It's the work I put off the longest and resent the most. You've probably seen me go a little quiet on platforms from time to time because, truth be told, feeding these platforms does my head in, lol.

But I keep the accounts alive, and the reason traces back to April of 2012.

That's when Facebook bought Instagram. Thirteen employees, about thirty million users, no revenue, and a headline price of a billion dollars. (By the time the deal actually closed that September, the stock had slid far enough that the real number landed somewhere around $715 million.)

That purchase is the reason Instagram isn't a photo app to me. It's an ad surface. It's the other half of the machine I run paid traffic through, and the grid is what people see when an ad catches them, and they click the profile to find out whether there's an actual human back there or just a landing page and a promise. An empty Instagram attached to a running ad is its own kind of answer, and it's not the one I want. So the account exists for continuity. It's the "yes, she's real, and yes, she's been at this a while" proof that makes the rest of the spend work.

Which means I need to post consistently on a platform I have almost zero desire to touch.

That's the tension I've been sitting with for a long time, and it's why building an Instagram carousel automation that genuinely runs without me got me so excited.

What the Instagram carousel automation actually does

It's a single scheduled task in Runable that runs every morning at 6 am, reads my Substack archive, picks a post, designs a carousel in my brand colors, writes the caption and hashtags, posts it to Instagram, and emails me when it runs.

That's the whole description. There's no second tool and no handoff.

I don't approve anything anymore, and I don't touch it, so the only part of this still in my calendar is reading the email over coffee.

I've been building automations for years now, from n8n workflows to agents to scheduled tasks running inside VS Code with Claude Code. None of that is required here, and that's actually the point I want to make. This one is different for a boring reason that turns out to matter a lot: I built the entire workflow and scheduled it in the same place, in about eight minutes. There was no creating something in one tool, exporting it, wiring it into a scheduler somewhere else, then debugging the handoff between the two.

You describe what you want, it does it once so you can look at the result, and then you hit schedule. That's it.

The setup, step by step

Four steps, and only one of them happens outside the chat window. Total build time was about eight minutes, most of which was me reading the sample output.

1. Connect Instagram

Go to Plugins, type Instagram, connect, grant permission. A modal pops up and handles it. This takes about thirty seconds, and it's the only "go do a thing in another window" moment in the entire build.

2. Feed it your brand

I already had my design system loaded into Runable as a skill file, which I'd originally built for Claude. That's the version of this I'd recommend if you have brand assets sitting around already... upload the design system once and stop pasting hex codes into every prompt forever.

If you don't have that, no problem. Put your colors, fonts, and tone directly in the prompt. It works either way, and the prompt below has a spot for it.

3. Point it at your archive, not your feed

This is the one decision I want you to actually think about, because it's the difference between a carousel account that stays interesting and one that recycles the same five posts forever.

You can point it at an RSS feed, and that's the obvious move. I pointed mine at my Substack archive instead, because the archive includes everything I've ever published, not just the last handful of posts. My content isn't gated (I think there are two old test posts behind a paywall and that's it), so the archive gives it a genuinely deep well to pull from.

If most of your good stuff is gated, the feed is the safer call. If it isn't, use the archive. You can point this at a website feed too, so it doesn't have to be Substack.

4. Write the prompt, check the sample, hit schedule

You paste the prompt, and Runable runs it once. Mine took seven minutes and eleven seconds, which sounds long until you remember what it's doing: choosing which tools to use, going out to the archive, reading the posts, picking one, designing the slides, and writing the caption.

The model selection is set to auto, so it picks for itself. You can override that and see token usage for each option if you want to be precious about it. I let it drive.

Then it hands you a finished carousel to review before you commit, and the Schedule button appears underneath the result. Pick your time, pick your notification, pick your start date, save.

Mine runs daily. You can set it to send you the carousel for approval first if you're not ready to let it post unsupervised, and honestly, that's the smart way to start. I ran mine on approval long enough to trust it and then turned it loose.

The prompt

Copy this, swap in your details, drop it into Runable. No opt-in, no gate.

Two notes on that:

Attach your logo file when you paste the prompt. You don't need to reference it by filename in the prompt text... it finds it.

And the slide count is a real variable, not a detail. I asked for six in the video and got five, because I'd edited the number in one place and not the other and didn't catch it. (I also said "or nine" out loud mid-sentence, so the confusion was entirely mine.) Decide on your number, put it in once, and check it before you schedule.

Questions you'll probably have

What I'm changing next

I want to test my Pixar character in these. She lives in a custom Gemini gem right now, and Runable isn't using Gemini for image generation, so the path is probably to upload reference images of her rather than expect a handoff between tools. That's the next experiment... followed by testing if Runable can create a Reel for me, based on my brand and character, then post that as well (my gut tells me it can before I even jump in to try it).

The part that actually matters

This didn't make me love social media, and I'm not going to stand here and tell you a tool changed how I feel about posting.

What it did was take a task I resent and take me out of it, while still producing something on-brand enough that when someone clicks through from an ad, the grid holds up. The Meta accounts do their job, and I do mine.

And notice what the tool didn't decide. It didn't know that my Instagram exists to support paid traffic rather than to grow a following. It didn't know to pull from the archive instead of the feed, that my content isn't gated so the archive was safe, or which posts translate to a carousel and which ones need the full read to land. Every one of those calls came from having run this business long enough to know what that account is actually for. The automation executed a strategy. It didn't have one.

That's the trade I've been trying to make for years, and the reason I kept not making it is that every previous version required me to maintain the automation itself, which is just a different job with the same amount of me in it. This one took about eight minutes to build and takes zero minutes a day to run.

What's the task in your business that you've been doing by hand because automating it felt like more work than just doing it?