Putting it together · for marketers
Building AI workflows in marketing
A single prompt can draft a blog post. A marketing team publishing every week needs more than that: research, an outline, a draft, a voice pass, a fact-check, and versions for each channel. Doing all of it in one prompt produces something mediocre at every step.
Putting it together means splitting recurring work into stages, choosing the right tool for each, and putting a human check where it matters. The result is a process the team can run again, not a lucky prompt.
Where it comes up
A content production pipeline
Topic research, outline, draft, edit, and channel versions, each as its own step. You review the outline before any drafting starts, because fixing direction late costs the most.
Monthly campaign reporting
Export channel data, have AI draft the summary against a fixed template, check every figure against the export, then write the recommendations yourself.
Repurposing one asset
Turn a webinar transcript into a blog post, three social posts, and a newsletter section. Each output needs its own instructions, drawn from the same checked source.
What it looks like at each band
E1 · Foundational
Uses AI for one step at a time and connects the steps by hand. Each new piece of content starts from a blank prompt. Results vary a lot from week to week.
E2 · Proficient
Runs recurring work as a sequence of prompts and reuses them. Saves good prompts and templates. Has a habit of reviewing at a fixed point before moving on.
E3 · Distinguished
Designs workflows with clear handoffs between AI steps and human review. Chooses different tools for different stages, such as one for research with sources and another for drafting. Documents the workflow so a teammate can run it.
E4 · Exceptional
Builds the team’s production system and measures it: where errors enter, which steps save time, and which steps need more review. Retires steps that no longer earn their place. Others adopt these workflows because they work.
Common mistakes
One giant prompt
Asking for research, a draft, and five channel versions at once gives the model too many jobs. Each stage gets less attention than it needs.
No checkpoint before drafting
If the outline or angle is wrong, every later step inherits the error. Review the direction before producing volume.
Automating what nobody has checked
A workflow that publishes or sends without a human reading the output will eventually ship an error at scale.
A worked example
Producing a weekly newsletter from the past week’s company blog posts and product updates.
Every Friday, paste all the week’s posts into a chat and ask for a newsletter.
The output changes format every week, sometimes misstates a product update, and has no fixed review point. Quality depends on how the chat happens to go that day.
Step 1: a saved prompt summarizes each post in two sentences with its link. Step 2: you choose the three items to lead with. Step 3: a saved template drafts the newsletter in the fixed sections. Step 4: you check each product statement against the release notes. Step 5: send.
Each step has one job. The human decisions, which items lead and whether the facts hold, sit at fixed points. The format stays consistent, and a teammate could run it from the saved prompts.
The full study guide for this skill, for every role, is Putting it together. The other five skills for marketers are on the AI skills for marketers page. A one-page summary is in the AI workflow checklist.
See where you stand
The free check grades putting it together and the other five skills in about fifteen minutes. Nobody sees the grade but you.
Then build the skills and prove it. The three steps for marketers list the courses and templates that fit the work.