I was watching an old episode of The West Wing recently, and there’s a scene where Toby Ziegler, the director of communications, is trying to get his secretary to build out an editorial calendar for the president’s campaign. She’s struggling to make the thing look right. Toby, being Toby, tells her it’s a calendar. Make it look like a calendar.
That episode aired more than twenty years ago, and I’d bet money that scene played out in your office last week. Somebody on your team is still trying to make a spreadsheet look like a calendar. And the reason it never quite works isn’t the formatting. It’s that a calendar of social posts is not really a calendar. It’s a caption, an asset, a link, an owner, a UTM, and an approval, all sitting in different places, pretending to be a grid.
I don’t code, and I don’t design, so for years my answer to that problem was the same as everyone else’s. Build the sheet. Link out to the folders. Send the sheet. And then wait.
The four-click approval
Claudia Sandino described the version of this that I think every social team lead will recognize. She sends her boss a link to the content calendar. He opens it. He clicks the link to the folder that has the creative. Then the link to the folder that has the copy. Then the link to the folder holding the tagged UTMs. Four destinations, and he still has to hold all of it in his head to approve one post.
Claudia has been doing this job for 15 years, for startups and for the Fortune 500, so this isn’t a beginner’s filing problem. It’s structural. Spreadsheets are good at keeping track of things and bad at showing you what a post will look like, and social is both of those jobs at once.
Her framing on the invisible part stuck with me. She puts 80% of a post behind the scenes and 20% on the platform. All that planning work is real work, and it produces an artifact nobody outside the team ever reads. As she said, your CEO does not care about your content calendar.
So the temptation, when a tool like Claude Design shows up, is to use it to make a prettier version of the artifact nobody reads. That’s the wrong move, and it’s the one I want to talk you out of.
Build something you can ask questions of
Here’s what changed for Claudia, and it isn’t the build time. The build time did change, from a week or a month down to a conversation, and that’s a real number worth having. But the thing that killed the four-click approval is that she stopped sending a document and started sending answers.
When her team asks for the partnership posts for the week, she doesn’t scroll to a tab and screenshot it. She tells Claude to pull all of the partnership posts for the week, create a separate document just for those, lay them out visually, and show the link, the caption and the visual. When her boss needs a narrower view, she asks for the next three posts, or the quarterly post, and sends him that.
That’s a different relationship with a deliverable. The calendar isn’t a file she maintains and distributes. It’s a source she queries, and the output is shaped to whoever asked.
How to build it, step by step
Load your brand before you prompt anything. Claudia goes to the design system panel first and drops in the brand bible: the logo, the fonts, the colors. You can also have Claude Code build the design system for you. Her warning here is worth repeating, because it runs against instinct: don’t over-Claude it. Bombarding it with every file you own produces worse output than giving it four things and refining as you go. If you’re on an enterprise account, your marketing team has likely uploaded this already, and anyone with access can use it.
Describe the calendar in one pass. This is close to what Claudia typed, and she was clear that she typed it without ceremony:
Think like a social media strategy director. Build me an interactive social media content calendar. Make sure to include dropdowns for Instagram, TikTok and LinkedIn. Include ownership tabs. Build it for the rest of the year, and because we’re social media, include key holidays.
Swap in your own platforms, and ask for the key holidays in your industry rather than the generic list. That’s where the National Pizza Day entries come from, if that’s still part of your programming.
Answer the intake questions. Claude Design came back asking about her role, her content pillars, her posting cadence, her industry, and whether she wanted the layout month by month. She answered social media manager, checked every pillar for the sake of the demo, chose four to five posts a week, picked the creator industry, and hit send. Her point about this step: you don’t have to have all the answers at the beginning.
Expect to run a second pass. The first draft came back, and Claudia said out loud that it looked clunky and wasn’t visually appealing. She asked it to improve the look and the branding, and it did. She calls herself a lazy prompter and defends it, because she’s having a conversation rather than writing a specification. I’d put it slightly differently. Her prompts can be loose because her judgment isn’t, and those are not the same skill.
Then edit by sentence. The moment in the recording that made me sit up was a scheduling change. Something happens in the world, you need to go dark, and the whole week has to move. Claudia typed one instruction:
Please move all of the posts from this week to next month.
The old version of that job was going into the sheet, moving everything, copying, pasting, reformatting, and hoping she hadn’t broken the format. She thinks it sounds ridiculous to worry about formatting. She also thinks the hours social managers pour into formatting calendars in Google Drive are insane. Both of those are true, and the second one is why this matters.
Share it as a live link, not a version. The share menu gives you an artifact: a PDF, HTML, PowerPoint, a PNG, or a video. Claudia copies the link and sends it, and her team sees the current state without her resending anything when it changes. She’s also honest that she has screenshotted a week, sent it to her boss, and gotten the approval that way. Not efficient. Worked anyway.
One detail here that I think is underrated: on an enterprise account, teammates can see her prompts and her edits. She reads that as a training channel, where a junior can watch how a senior actually built the calendar. Most tools hide the reasoning and show only the artifact. This one leaves the working out on the page.
Where it breaks
I won’t pretend this is finished technology, and Claudia didn’t either.
Claude Design isn’t multiplayer. Several people can’t sit in the file and edit it together, which is a real gap when your calendar is a team object. Her workaround is a handoff to Canva through MCP, where the team can collaborate properly. If you want the posts scheduled after that, MCP also connects Claude to Agorapulse, so the assets and the calendar can move from Design to Canva to a publishing queue without a manual export in the middle.
The bigger limitation is accuracy. Working raw, with no brand kit and no MCP connected, Claudia’s assessment is blunt: it just makes things up. So she reviews everything. She reads every single cell. She confirms the dates are correct, that the dates align with the month, and that the holidays are the same. She said 100% about that twice, and she meant it as a floor rather than a flourish.
The third constraint isn’t the tool’s fault. You can’t come to Claude Design and ask it to do the job for you. You have to do the work first, which for her means the brainstorm happens in Claude before the build happens in Design. If you’re starting from scratch or you don’t know what the calendar needs yet, go there first. If you already know, skip it and start building. We compared notes on this during the episode and landed in different places, which I think is the honest answer: the method depends on whether the spec is already in your head.
If you want the full walkthrough, including her screen and every prompt in sequence, it’s all in the episode breakdown.
The part that outlasts the tool
Strip away Claude Design for a second, because in eighteen months there will be three more tools doing this and one of them will do it better.
What’s actually changed is the nature of the deliverable. For twenty years, our planning artifacts have been documents: things we build, name, version, store, and circulate. The work of maintaining them was invisible, and the work of reading them fell on whoever we sent them to. Toby’s secretary was fighting that. So were you, last Tuesday.
A queryable artifact moves that burden. Nobody clicks through four folders because nobody has to navigate anything. They ask for what they need and get exactly that view. Your calendar stops being a document about the work and starts being an interface to it.
So when you evaluate the next tool that lands in your feed, I’d skip the demo reel and ask one question: can the output answer a question, or does it just look better than what I have? A prettier version of a document nobody reads is still a document nobody reads.
And keep Claudia’s constraint in view, because it’s the one people will ignore. Her sloppy prompts work because 15 years of building these calendars by hand told her what to ask for and what to check. The tool supplied the speed. She supplied the judgment. When you build your first one, budget real time for that review pass, and read every cell.
