What a multi-channel content pack is, and why one brief should produce all of it
A multi-channel content pack is a single brief turned into every asset a campaign needs. In one run it produces a blog draft, a post written for each social platform, the email copy, and the image direction from one argument, so the pieces agree with each other instead of drifting apart.
The real failure mode is drift. If four different people write four separate assets on four different days, each from an empty document, the blog makes one statement, the email makes a neighboring one, and the ad makes a third. Nobody notices until the campaign goes live, at which point the landing page quietly contradicts the post that directed people to it.
The fix is structural rather than editorial. The argument is prepared once, and then each channel is derived from it, so consistency is a feature of how the materials were created rather than something the reviewer has to catch at the end.
With ActionFlows, this flow is based on the Generate Object node, which returns all the channels within a single validated structure rather than as four separate generations.
How the ActionFlow runs on the canvas
The Start node takes the brief as structured input: the audience, the main message, the channels on which you publish, and any constraint that matters (a character limit, a claim that has already been cleared, or a product name that must be used exactly as it appears). Start Scheduled is used when the pack is produced on a set schedule rather than in response to a request.
The first Generate Text node is built not to produce anything that could be published. It sets out the main argument in three sentences. Every later node takes its input from that output rather than from the original brief.
The Generate Object step provides the structure. Rather than generating a separate text for each channel, it returns a single structured object with a separate field for each channel (blog, x, linkedin, email, and image_prompt) according to a schema you set. Four separate generations can quietly differ about the offer. These fields cannot, because they are generated together, from the same argument, in the same call. A Validate node follows. If any required field is empty, it stops the run.
The image_prompt field feeds into Generate Image, so the visual is generated from the argument rather than chosen afterward from a stock library.
Distribution is kept separate on purpose. A Notion node writes the finished pack as a single page, with the core argument at the top. A Slack node in the Human in the Loop category holds the pack until a person approves. Approval is a node inside the graph, not a step someone remembers. Only after approval does an X (Twitter) node publish.
Nodes this ActionFlow uses
- Start: receives the brief as structured input. Start Scheduled produces packs on a cadence.
- Generate Text (AI Core): writes the core argument in three sentences.
- Generate Object (AI Core): returns one object with a field per channel, shaped by your schema.
- Validate (Data): stops the run when a required channel field comes back empty.
- Generate Image (AI Core): takes the image_prompt field so the visual carries the same argument.
- Notion (Integration): writes the pack as one page with the argument at the top.
- Slack (Human in the Loop): holds the pack for approval inside the graph.
- X (Twitter) (Integration): publishes only after approval.
Frequently asked questions
Start building AI workflows
Create a free account, open a template or a blank canvas, and run your first ActionFlow.
Newsletter
Get product updates
New nodes, agents, and product notes. We send mail only when we have something worth opening.