I get a lot of questions about my content engine. How does one video become an article, graphics and posts for different channels without everything sounding like the same generic summary?

In this walkthrough, I show you the part of my process that starts with a finished recording. I give the agent a transcript, then work through the decisions and reviews that turn it into a content package. You’ll see ChatGPT, my saved writing instructions, Airtable, Figma and the finished materials together.

The transcript supplies the material. The instructions tell the agent what to do with it.

My workflow references the way I want to write for each channel, the examples I want it to learn from and the recipe for each type of output. When I review content performance, I feed what I’ve learned back into those references. Over time, that gives the agent a more specific starting point than a fresh request to write a LinkedIn post or a newsletter.

The four steps here begin with a finished recording and transcript. In the larger workflow, that’s Step 5: the research and recording have already happened. To try this with your team, start with one transcript, a few approved writing examples and a clear description of the output you need. Give the agent those references before asking it to draft.

One of the most important stops comes before the writing.

I’ve asked the agent to recommend an angle and offer alternatives. In the example, the source video is about explaining “on brand” to a design agent. It suggests a few ways into that topic, and I choose one with a qualification from another.

That decision gives the work a point of view. Otherwise, a lot of AI-written content can feel flat and without purpose, even when it covers the source material. I’d rather make the editorial decision here than ask for revisions across a whole batch later.

The next stop is the newsletter itself. I review the copy, subject and preview text before asking the agent to make the visuals. For your first run, ask for a recommended angle and alternatives, then choose the point you want readers to take away. Review one complete draft before turning it into a batch of channel-specific pieces. You can change an argument much more easily while it’s still one article.

My Figma library contains the components and examples I want the agent to use. It also has an introduction explaining what’s in the file, how to use it and where to find things.

That matters because a brand palette alone leaves a lot for the agent to invent. A finished editorial card, carousel or thumbnail gives it a more concrete reference. In the demo, the agent uses the library, takes screenshots and compares its work with the examples.

I still review the result. Sometimes I want another graphic, or I want it to illustrate a different part of the article. I can ask for that change in the conversation or edit the design directly in Figma. Start with one format your team already uses, such as an editorial card or carousel. Give the agent the actual design reference and ask it to compare the result with that example. Check the text, spacing and crop before asking for more versions.

The finished materials need to be available to people beyond the person running the chat.

My workflow saves the media to Drive and checks the downloaded files against the originals. Airtable tracks the finished work and its review status. At the end of the demo, I ask it to save the package for final review before anything goes out to the platforms.

This is more involved than working alone and making a few pieces for yourself. Most of the clients I work with have multiple people who need to see, refine and approve the work. They also need the graphics to stay consistent from one run to the next. The saved references and handoff are there to support that way of working.

For your first handoff, put the article, channel copy and graphics somewhere the next reviewer can open them. Make it clear which version they’re reviewing and what still needs a decision. Check the files and links from that person’s point of view. If they have to reconstruct the work from your chat, the handoff isn’t finished.

You can explore the shared workflows in our AI Marketing OS repository on GitHub. Start with its README and choose one workflow to try with your own source material. You’ll still need to supply your team’s context, writing examples and review decisions.

Thanks for reading. I hope this gave you something to bring back to your team.

At Multipl.ai Growth, we’ve built AIMOS, an agentic marketing harness available via MCP. It gives AI assistants a shared set of marketing skills and workflows, from customer research and positioning to content, SEO/AEO, campaign planning, creative production, and paid media audits.

We’re also a consulting firm that embeds an AI-native marketing leader in your team to build a custom AI marketing harness for your business. We bring the latest methods, technologies, and multiplayer AI frameworks into the work, connecting your business context, tools, and workflows so your team can build and work with AI together.

If you’d like to explore what that could look like for your team, book a 20-minute discovery call here. We’d love to hear what you’re working on.