A guide to creating an AI-powered content idea assistant
A quick-start guide based on How Joy Joya went from 10 hours to 1 of content brainstorming per week, featuring Laryssa Wirstiuk, published on the AI Lab by ActiveCampaign.

Unlock the quick-start guide
A quick-start guide based on How Joy Joya went from 10 hours to 1 of content brainstorming per week, featuring Laryssa Wirstiuk and published on the AI Lab by ActiveCampaign
What you’ll build
With this workflow, your team can request AI-generated, on-brand social media content ideas directly from Slack. The AI pulls brand context from Notion, generates 10–15 tailored ideas, and organizes them in a Notion database for your team to review. This quick-start guide gets you a working prototype for one client that you can then expand to multiple clients later.
Time commitment
- Quick version (this guide): 2–3 hours
- Full version (see original post): 2–4 weeks (multi-client, with developer, approval-to-calendar automation)
What you need
- Slack workspace with a dedicated channel (free plan works)
- Make account (free tier allows 1,000 operations/month — enough for testing)
- Notion account with a workspace for client documentation
- OpenAI API key (ChatGPT access)
The build (minimum viable version)
Step 1: Create the brand profile in Notion — 30 minutes
Create a single Notion page for your first client with these fields: brand voice (2–3 sentences describing tone), content pillars (3–5 themes), product categories, and any current promotions. This is what the AI uses to make ideas specific rather than generic.
Quick version: Don’t overthink the brand profile. Copy-paste from an existing brand brief or style guide. You can refine it after seeing the first batch of AI output.
Step 2: Set up the Notion review database — 15 minutes
Create a Notion database called “Content Idea Review” with these properties: Title (text), Content Type (select: storytelling, educational, product highlight, behind-the-scenes, seasonal), Description (text), Status (select: In Review, Approved, Rejected, Transferred), Client (text), Week (date).
Quick version: Use Notion’s “Board” view grouped by Status so your team can drag ideas between columns during review.
Step 3: Build the Make scenario — 60–90 minutes
Create a new scenario in Make with this sequence:
- Slack trigger — “Watch New Messages” module pointed at your dedicated channel
- OpenAI module — Send a prompt that includes the Slack message text and your client’s brand context (copy-paste the Notion brand profile into the prompt for now)
- Text parser — Split the AI response into individual ideas
- Notion module — “Create a Database Item” for each idea with status “In Review”
Quick version: Skip the dynamic Notion lookup for now. Hard-code your first client’s brand context directly in the Make prompt. You can add the Notion retrieval step later when you expand to multiple clients.
Step 4: Test with a real request — 15 minutes
Post a message in your Slack channel: “Generate ideas for [Client Name], [next Monday’s date].” Watch the Make scenario execute. Check your Notion database for the new ideas. Verify they reference the client’s brand voice and current promotions.
Quick version: If the ideas feel generic, the brand profile needs more detail. Add 2–3 example past posts that captured the brand well.
You’re done when…
- You can post a message in Slack and see 10–15 content ideas appear in your Notion database within a few minutes
- The ideas reference your client’s specific brand voice, promotions, and content pillars
- Your team can review, edit, and mark ideas as “Approved” or “Rejected” in Notion
Expand later
- Add dynamic Notion lookup so Make pulls brand context automatically per client
- Build the second Make scenario that moves “Approved” ideas to the content calendar
- Add the cleanup AI step (second-person voice, platform-agnostic language, consistent formatting)
- Add routers in Make to handle multiple clients with different calendar structures
- Add break modules to prevent Notion API throttling at scale
- Hire a Make developer (Upwork) to architect conditional logic and edge case handling
Ready for the full story?
Read How Joy Joya went from 10 hours to 1 of content brainstorming per week, featuring Laryssa Wirstiuk, published on the AI Lab by ActiveCampaign.
If you’re ready to build your own:
Related
More data from the AI Lab.