When should you use it?
- A team writes new ad concepts every week and runs out of fresh angles
- You sell into a market where customers talk openly in forums, reviews or communities
Not for: Copying competitors' ads; the point is to find what they're not saying
What do you need, and what do you get?
You need
- Where your customers talk (subreddits, review sites, forums, comment sections)
- Competitor names for public ad libraries such as Meta's
- A Slack channel or email for the report
You get
- New angles and the customer wording behind them, with links
- What competitors' current ads emphasize
- Gaps: needs customers raise that no ad addresses
How does it work, step by step?
- List the sourcesWrite down where your buyers talk and which competitors to watch; keep it to a handful at first.
- Write the brief onceUse our prompt as the automation's instructions, with your product and audience filled in.
- Schedule and deliverRun it daily or weekly and post the result to one Slack channel.
- Feed back resultsTell it which suggested angles became winning ads so later reports lean that way.
The prompt
Our own prompt, inspired by Camm (@cjfeth) and written to work in Manus, ChatGPT agent and Claude. Values in lime are placeholders.
Fill in the blanks (optional; the copied prompt uses your values)
Research brief for {product}, sold to {audience}. 1. Read the newest public posts in {customer_sources}. Pull out what people want, complain about, have already tried and rejected, and the exact words they use. Link each point. 2. Read the active ads for {competitors} in public ad libraries. Summarize each brand's main promise, hooks and offers. 3. Compare the two. List up to 5 gaps: needs customers raise that no competitor ad speaks to, with the supporting customer posts. 4. For each gap, suggest one ad angle in a sentence. Don't copy any competitor wording. Keep it scannable for a Slack post.
Inspired by Camm (@cjfeth): Manus automation for creative strategy research. The prompt above is written by playbooks.bot in our own words; nothing from the original is quoted.
What can go wrong?
- Ad libraries show what is running, not what is working. Treat competitor ads as a map, not proof.
- Quote customers only from public posts and never copy competitor creative.
Our take
Ad creative usually stalls on research, not ideas: someone has to read what customers complain about and what competitors already claim before anyone can spot an opening. That reading is repetitive and easy to hand off, and Camm's setup does it in the right order. It starts from the customer's own words, sorts them into angles and objections, and only then compares against competitor ads, so the output is gaps rather than imitation. Two cautions. Ad libraries show what is running, not what is profitable, so a crowded angle may be crowded because it works. And a daily report can turn into noise; ask for at most five gaps and include links so strategists can read the source threads themselves. In our dry run with made-up forum posts about a meal-kit service and copy from three made-up competitor ads, our prompt found that all three ads led with price while customers mostly discussed food waste and picky kids, and proposed three angles with the customer quotes and links behind them.
Where the time goes back: The research your strategists do before each brainstorm arrives in Slack each morning, with sources attached.
FAQ
How often should it run?
Which agent can do this?
Source and credit
Inspired by Camm (@cjfeth). We describe the idea in our own words, link to the original, and wrote the prompt ourselves; nothing is quoted.
- Inspired by
- Manus automation for creative strategy research
- Creator
- Camm (@cjfeth)
- Published on
- X
- Original date
- Mar 18, 2026
- License
- Not openly licensed (the poster keeps all rights), so we quote nothing and link to the original
- Originally for
- Manus
- AI dry run
- Oct 20, 2026: passed. Dry-run tested by our AI test agent on sample input. Output was gaps, not imitation, and every point had a source.
AI dry run on sample data; not run with live connected accounts. The playbooks.bot test agent filled the prompt with synthetic inputs, produced the output and checked it against the page's stated outputs and guardrails. - Verified
- Oct 23, 2026: source link resolves to the original, the credited creator published it, our excerpt or prompt matches it, and the license allows our use.
Are you the creator? Claim or correct this page.